The following is a special guest post by Jon Miller, a legend in our industry, the co-founder and original CMO of Marketo, and a leading architect of modern B2B marketing and ABM as we know it. Now the co-founder and CEO of a new stealth AI startup in this space, this is what he anticipates we’ll see in the year ahead.

This is my fourth year writing predictions for B2B go-to-market. You can read the 2023, 2024, and 2025 editions, and I publicly grade myself each year. (Here are my grades for 2025.) I don’t always get it right, but forcing myself to commit to specific predictions — and then facing the scorecard — helps me think about what’s actually changing (versus what’s just noise), and how to navigate the year.

2026 feels like a year where the vision of AI-powered marketing comes into focus: agents joining buying committees, reasoning systems replacing brittle rules, orchestration that finally delivers on the decades-old promise of 1:1 personalization. But actual enterprise adoption will remain incremental and the gap between where we’re going and what actually happens in 2026 will be large.

Here are the predictions:

1. Marketers will begin marketing to agents, not just humans
2. AI will completely transform legacy SaaS martech — but not in 2026
3. Composable stacks will be mainstream by 2030, but <20% will adopt in 2026
4. Context engineering will emerge as a recognized practice across GTM teams
5. Reasoning AI will begin replacing rules-based automation
6. Journey orchestration shifts from rules to AI playlists, delivering on the 1:1 promise
7. AI inbox gatekeepers will turn email marketing into earned media
8. Taste, trust, and accountability will become the antidote to AI slop
9. Public intent signals will commoditize; proprietary signals generate “alpha”
10. Where martech is heading beyond 2026
11. Rapid-fire mini-predictions
12. AI-driven uncertainty will intensify in 2026; preparation is the only answer

Marketers will begin marketing to agents, not just humans

Prediction: In 2026, marketers will market to agents, not just humans. As buying journeys become increasingly AI-mediated — with agents researching and evaluating vendors alongside human stakeholders — marketers will deliver the structured information AI needs, while also investing in the human experiences and relationships that close complex deals. Scott Brinker calls this ‘from martech to marketing to tech’, treating AI agents as part of the buying committee.

What’s already happened:

2025 was the year of Answer Engine Optimization (AEO). 90% of B2B buyers used tools like ChatGPT for research, and 72% encountered Google AI Overviews during vendor evaluation (TrustRadius 2025 report). That’s why 63% of marketers are publishing AI-optimized content like structured FAQs and schema markup (Martech for 2026 report from Chiefmartec and MartechTribe), as well as optimizing their presence on platforms where LLMs source answers like Reddit and G2. And the Digital Bloom reports 51% of companies plan to increase AEO investment, versus 14% for traditional SEO.

What will change:

AI becomes the new analyst briefing: How AI describes your company will be a key product marketing KPI. Executive teams will obsess over how AI positions them versus competitors, the same way they obsessed over Google rankings. 

AI advertising: We will see early adoption of paid ads in AI results. Perplexity already has sponsored follow-up questions and Google shows search and shopping ads in AI-generated summaries, especially for complex or commercial queries. OpenAI has plans to monetize free users with ads, though those are temporarily deprioritized following December’s “code red” to focus on core product quality. I’m especially excited to see audience targeting come to paid AEO, though I suspect this is post-2026. The ability to bid more for known contacts and accounts — similar to how paid search can target specific audiences today — will let marketers focus on the buyers and accounts that matter most.

Agent identification and tracking: When an AI agent requests pricing information or product specs from your site, that’s first-party intent, and teams should incorporate those signals into scoring and buying stage predictions. Put simply, agent visitors will contribute to Marketing Qualified Accounts (MQAs) just like human visitors. This will require “agent deanonymization”: an unknown startup’s agent researching your category may be noise, while a Fortune 500 company’s agent comparing your capabilities against competitors could be pipeline.

Content architecture for dual audiences: Marketing to both humans and agents will require rethinking content architecture. This will lead to a “headless” model: a foundation layer of core data, content, and services that serves two distribution paths. Humans get branded website experiences optimized for emotional engagement. Agents get direct access to services — think pricing APIs that deliver dynamic quotes, or knowledge bases that can answer detailed product questions. Both get responses personalized to their role and buying stage. Building this capability requires robust infrastructure that can field detailed queries accurately; chat vendors like Qualified have been building these systems, which likely contributed to their acquisition by Salesforce.

Agent nurturing: This one is pure speculation, but it may be computationally expensive for an agent to regularly ping a company for relevant updates. Agents might instead ‘register’ or ‘opt-in’ to receive proactive updates from companies their humans care about. If that happens, ‘agent nurturing’ could emerge as a channel.

What won’t change:

AI agents optimize for objective criteria: capabilities, pricing, compliance requirements. They don’t care about your brand story.

But humans still make final decisions, at least for now. The 6sense Buyer Experience Report for 2025 showed human buyers averaged 16 vendor interactions, unchanged from 2024. Complex purchases require human validation, relationship building, and trust. Agents compress research time, but people sign contracts.

Hot take: AEO tactics will prove temporary

The work to optimize static content for AI engines (special markup, structured FAQs, machine-readable formatting) feels like early SEO, when people stuffed keywords and gamed algorithms. Google has long said don’t create content for machines, create for humans. Even their current AI documentation confirms: “You don’t need to create new machine-readable files, AI text files, or markup to appear in these features.”

The headless service infrastructure I described above may prove durable, but the static content markup tactics will likely become unnecessary as AI gets better at reading human-optimized content. Over time, I predict we won’t need to optimize our content for AI, which is exactly what happened in search.

AI will completely transform legacy SaaS martech — but not in 2026

If you spend any time on LinkedIn, you’ve seen the proclamation: “SaaS is dead.” As Scott Brinker says, if you only follow the techno-optimist influencers claiming to run entire departments with vibe-coded agents, you might feel desperately behind.

Here’s what that vision looks like: you describe your goal and AI orchestrates everything across channels, no campaign builder required. “Promote our Q4 launch to enterprise accounts in financial services” becomes a single request that an AI translates into personalized content, coordinated timing, channel selection, and budget optimization.

Prediction: Yes, AI will reshape marketing technology — but it will take the next 2-5 years. 2026 will be a year of hybrid experimentation, not wholesale transformation, in which SaaS platforms (old and new) will coexist with agentic AI, and human-in-the-loop will remain essential across nearly all implementations.

The three disruption fronts:

AI is driving three disruptive shifts in legacy SaaS:

1. From software tools to autonomous agents: When you can hire digital workers who never sleep, never quit, and scale infinitely, why pay for software licenses? Many of the fastest-growing AI companies (Cursor, Harvey, etc.) aren’t about helping humans work better, they’re about replacing labor entirely. 

Brendan Short argues the next major GTM company will be one that “sells labor” rather than software, likely in the form of AI agents managed by GTM operators. You’ll no longer buy software that helps people do work; you’ll hire AI to do the work. You’re paying for outcomes, not seats.

2. From user interfaces to headless applications: Instead of logging into individual platforms, many humans will work in their chatbot of choice. For these users, the functionality exists, but the user interface layer disappears. A great use case for this is analytics; a CMO simply asks their AI chatbot “which campaigns performed best this quarter”? (That said, I’ll note that chat isn’t always better than clicks. For experienced users doing repetitive tasks, a familiar UI often beats describing what they want and waiting for an agent to interpret it.)

This is why the Model Context Protocol (MCP) is so exciting. And MCP solves the integration mess: instead of building custom connectors between every AI application and every tool, it provides a standardized protocol where each builds one connection and everything works together. 

3. From monolithic platforms to composable apps: AI can now code mini-applications to solve specific problems, often without being asked. And users can “vibe code” simple solutions themselves. Over time, this will reduce the number of lower-value “apps” in various app stores. 

But there are limits to how far this can go. “Martech for 2026” shared a good framework: simple apps (basic dashboards, internal tracking tools) work fine for non-engineers. But anything handling sensitive data needs qualified developers, and complex enterprise platforms (CRM, MAP) should still be bought from commercial vendors. The likely outcome: core platforms from trusted vendors, surrounded by bespoke mini-apps filling company-specific needs.

Why not 2026:

These disruptions will transform martech, but more slowly than the hype suggests. In 2025, 81% of AI usage is ‘assist-only’ while less than 10% let agents act autonomously. Why so slow?

• Enterprises move slowly. Brinker’s Martec’s Law states that technology changes faster than organizations can absorb it. Think how long past tech transitions (mainframe to enterprise software, enterprise software to SaaS) took. Within marketing, consider how long MQLs have persisted despite better options like MQAs. The shift is happening, but it’s measured in years, not quarters.
  
• Enterprise requirements. SaaS platforms are more than their user interfaces. They’re valued for the infrastructure that takes years to build right: compliance and security, governance and auditability, scalability and integrations. AI doesn’t eliminate the need for these capabilities.
  
• Technical complexity. Dynamic agent orchestration for millions of contacts with complex real-time decisioning requires serious infrastructure. Bidirectional sync with Salesforce, complex data models, edge case handling — these hard engineering problems don’t disappear because you added an AI layer.
  
• AI hallucinations persist. The risk of confident but incorrect outputs remains a significant barrier to autonomous operation, particularly for mission-critical activities. You can’t have agents dynamically deciding whether to honor CAN-SPAM opt-outs or GDPR deletion requests. So human-in-the-loop will be around for a while.
  
• The pricing model tells the real story. If SaaS were truly dead, we’d see widespread outcome-based pricing. Instead, Salesforce recently introduced its Agentic Enterprise License Agreement, which uses credits but is based on seats at the core. As CFO Amy Weaver explained, customers want seat-based pricing because ‘it gives you predictability.’ That tells you where enterprise confidence actually is.

What 2026 will actually look like:

Given these challenges, 2026 won’t look like autonomous agents running headless applications. It will look like SaaS with AI built in. Think of it like self-driving cars: we’re not getting full autonomy in 2026. We’re getting advanced driver assistance where the human stays alert and in control, ready to take over.

In 2026, enterprise marketers want a better Marketo, not a wholesale re-architecture. They want AI-assisted campaign creation that still shows them the workflow before it executes. They want intelligent audience segmentation they can review and adjust. They want AI providing “air traffic control” between campaigns to prevent message conflicts, but with traditional interfaces and human approval loops. That’s what will actually get adopted.

This also explains why we’ll likely see more traction from new AI-native platforms than from AI add-ons trying to retrofit legacy systems. The new platforms will be architected to be agentic, API-driven, and MCP-enabled at their core, while still supporting today’s UI-based workflows and human oversight patterns. Think of them as building the bridge: functional today with traditional interfaces, but ready to support fully agentic operation as enterprises become comfortable with that shift.

Composable stacks will be mainstream by 2030, but <20% will adopt in 2026

A composable martech stack is built from modular layers rather than a single vendor suite:

• A Data layer serves as the source of truth, combining customer data from multiple sources into actionable inputs
• A Decisioning layer handles intelligence like audience selection and journey orchestration
• Various Delivery tools execute messages across channels via APIs

The idea is that you choose the best tool for each job, swap components when better options emerge, and avoid vendor lock-in. 

B2B marketers are moving in this direction, though slowly. According to The Digital Bloom’s analysis of martech stacks, marketing automation platforms’ role as the “center” of B2B stacks declined from ~31% to ~26% of respondents year-over-year, while custom platforms (a proxy for composable) grew from 2% to 10%. Today, most mid-market B2B firms still favor integrated platforms (like Marketo or HubSpot) over assembling a complex stack, but the trend is clear.

Prediction: By 2030, a modular, AI-native, signal-driven stack will be the norm in B2B marketing. But fewer than 20% of B2B teams will run a fully composable architecture in 2026.

Why so slow?

Flexibility shifts complexity; it doesn’t remove it. A composable stack trades vendor-managed convenience for flexibility and control. You gain customization, but you assume responsibility for making it all work together. MCP will slowly make this better.

Multi-vendor management is harder than it looks. Coordinating updates, support, contracts, and roadmaps across multiple vendors requires significant operational maturity. As Mike Lowndes, VP Analyst at Gartner, noted: “Especially in B2B, the challenges of adopting composable solutions are significant. Many organizations struggle to manage multiple vendors and contracts effectively, particularly if they lack the necessary digital maturity.”

Data warehouses aren’t built for marketers (yet). The idea of a central warehouse fueling all marketing sounds great, but in my experience these platforms are typically built for engineers and analysts, not everyday marketing users. Legacy martech platforms struggle to access the product or behavioral data stored there, leading to data silos, API limit workarounds, and engineers manually extracting data for marketing.

Decisioning and Delivery are hard to separate. In theory, an AI model in a standalone tool decides the next-best action and then triggers execution via API. In practice, it’s harder than it sounds. Your ESP might not support the dynamic personalization your model envisioned. More importantly, truly optimizing outcomes requires a closed feedback loop: capturing response data, converting it to learning, and feeding it back. As Real Story Group observed, “true orchestration requires closing the loop across data, decisioning, and content layers” — something few stacks achieve today.

The likely path: composable lite

Most B2B companies won’t leap to fully headless architecture overnight. Instead, they’ll adopt what I’d call “composable lite”: data-first and decision-first, but not fully decoupled.

In this model, the data warehouse becomes an important source of truth, but your MAP, website, chat tools, ad platforms, etc. aren’t just dumb execution channels. Final journey orchestration will live inside the MAP, even as scoring, segmentation, and some intelligence shifts toward the warehouse. It’s a stepping stone that lets teams capture real benefits without dismantling everything at once.

Context engineering will emerge as a recognized practice across GTM teams

Poor data quality is consistently cited as the biggest obstacle teams face with AI systems. But “data quality” undersells the problem. The real gap is giving AI the operational context it needs to act like someone who actually understands your business.

Prediction: In 2026, “context engineering” — the discipline of systematically capturing and structuring the knowledge that makes AI useful rather than generic — will emerge as a recognized practice across GTM operations teams.

What counts as context? It starts with connecting AI to the same data and knowledge sources your team relies on: CRM, data warehouse, call recordings, marketing platforms, internal documentation. But connection alone isn’t enough. AI also needs an enablement layer that explains what the data means and why. That means naming conventions for campaigns. The rules for which segments to include or exclude, and the reasoning behind those rules. What a ‘webinar campaign’ means at your company: which tactics to create, what the follow-up sequence looks like, how leads get routed. The reasoning behind your Salesforce schema after ten years of accumulated customization.

This operational knowledge isn’t formally captured anywhere. It lives in Slack threads, in tribal knowledge, in the heads of senior team members. When your best MOPs person leaves, it walks out the door with them.

Legacy marketing platforms don’t help. They can tell you what campaign a lead is in. They can’t tell an AI agent why that campaign exists, what the underlying strategy was, or which approaches your team tried and abandoned. Legacy systems store outcomes; they don’t capture decision logic.

What changes:

• Ops teams will spend time teaching their AI platforms ‘skills’: how to interpret the data schema; how to build a webinar campaign; which segments to use, and when. This looks a lot like onboarding a new hire, except the knowledge becomes durable and reusable rather than trapped in someone’s head. It also helps Ops evolve from a tactical ticket desk, becoming the team strategically responsible for making AI actually useful.
  
• AI systems will start capturing decision outcomes (what worked, what didn’t, why) so the next decision builds on the last. That’s the difference between AI that repeats mistakes and AI that learns how your business operates.
  
• As teams deploy multiple AI agents, the shared operational context becomes the coordination layer. Without it, agents work at cross-purposes; with it, they act like a coherent team.
  
• Martech vendors will begin to compete on how well they capture and expose operational context, not just on integrations or features.

Reasoning AI will begin replacing rules-based automation

Legacy marketing technology is built on rules engines. If job title contains “VP,” add 10 points. If email opened, wait two days, then send follow-up. If industry equals “Financial Services,” route to segment B.

Those rules are brittle. They can’t learn from outcomes, they break when market conditions shift, and they require expert-level knowledge and constant maintenance. Worse, they can’t handle the ambiguity that defines the real world.

Reasoning models can. They reason through problems, understand context, and test hypotheses. They recognize patterns without explicit rules, infer relationships from available data, and weigh multiple signals simultaneously.

Prediction: In 2026, reasoning AI will begin replacing rules-based logic across marketing and revenue operations, starting with data management, lead scoring, and journey orchestration (see next prediction).

The “Martech for 2026” report (Brinker, Riemersma) offers a useful spectrum for automation, from deterministic (repeatable and explainable, but rigid) to agentic (adaptive and resilient, but less predictable). Most legacy marketing tech sits on the deterministic end. The opportunity in 2026 is to move selectively toward the middle, incorporating reasoning AI for automations that benefit from contextual judgment while keeping deterministic logic where consistency matters.

This shift addresses what Justin Norris calls the “messy middle”: work that’s too variable for rigid rules but not strategic enough to justify senior attention. The Slack pings, data cleanups, exception handling, and small fixes that bury ops teams. 

Marketing ops won’t disappear. Instead of configuring complex rule systems, teams will provide context: business goals, success metrics, guardrails, and data pipelines that give AI access to the signals it needs.

Journey orchestration shifts from rules to AI playlists, delivering on the 1:1 promise

The promise of 1:1 personalization has been around for decades, yet it remains mostly unfulfilled in B2B. After years of inflated vendor claims, any pitch about “personalized journeys” is met with justified skepticism.

Why has it been so hard? B2B buying involves non-linear journeys and large buying committees. Rules-based personalization devolves into spaghetti diagrams that are theoretically possible but practically unmanageable. And “next-best-action” is too simplistic for long B2B journeys; we need to think multiple moves ahead, not just one.

Prediction: In 2026, early adopters will implement AI-powered journey orchestration that dynamically sequences actions based on real-time signals. This helps AI drive more effectiveness, not just efficiency.

Why now?

This is a specific case of the shift from rules to reasoning described in the previous prediction. Rather than mapping every possibility into a complex workflow diagram, AI can finally think through all the possibilities to pick the best path for each buyer and account.

But journey orchestration also benefits from a second breakthrough: modern AI’s ability to work with the kind of data B2B actually has.

Traditional machine learning tended to flatten the behavioral signals that matter most, like which specific web pages and campaigns a buyer engaged with. Transformers and large language models work differently. They can encode everything known about a buyer and their account — engagement history, content preferences, timing patterns — into rich representations that capture behavioral nuances, making them perfect for B2B.

Right action beats content personalization

Most conversations about personalization focus on content: AI-generated emails mentioning someone’s LinkedIn post, dynamic copy blocks, personalized subject lines. 

I think that’s the wrong focus. To me, 1:1 personalization means figuring out the right action for every buyer and account. An action combines four elements: offer, channel, content, and timing. You don’t need unique content for every buyer. You need the right action at the right time.

Instagram and TikTok prove this at scale. They don’t create unique content for each user; they intelligently sequence existing content into feeds that feel deeply personal. 

Playlists, not shuffle

Most personalization engines pick the single best next action. That’s like putting music on shuffle: each song might be fine, but you lose the artistry of a well-sequenced album. In B2B, nobody buys because of one touch. And just as a language model writes better by looking several tokens ahead, AI can create better journeys by looking several actions ahead.

I call this approach “Playlists”: a sequence of upcoming actions, dynamically adjusted based on engagement signals and buyer context. You set the strategy and goals (the genre or mood), and AI curates a personalized sequence to move each buyer forward.

Playlists support:

• Dynamic sequencing: The playlist adjusts in real time based on engagement signals, timing, and context rather than following a fixed path.
• Mix and match at scale: Combining offers, channels, content, and timing from pre-approved libraries creates billions of unique action sequences without requiring millions of custom assets.
• Reinforcement learning: Instagram doesn’t know you like cat videos until it shows you one. You don’t know an executive prefers Sunday emails until you try sending on Sunday. The system explores, learns, and adapts.

Changing roles

In this model, the human’s job is to create compelling offers, experiences, and content: understanding markets, developing messages that resonate, crafting experiences that connect. The human also sets the boundaries: rules about who must or must not receive certain communications, budget limits, frequency caps, brand guardrails. They are defining the playlist’s constraints, not programming every song.

The AI figures out who gets what and when, across dozens of campaigns and thousands of accounts. It also handles air traffic control, ensuring buyers aren’t overwhelmed with conflicting messages. Humans shouldn’t be building Visio diagrams with endless if-then logic; we’re terrible at that complexity. Let AI handle the combinatorial math.

Meanwhile, humans will still monitor AI recommendations and make final decisions for a while. And transparency will always be non-negotiable. Why did this person get this campaign? Why didn’t someone qualify? What alternatives were considered? Without answers, teams can’t learn, debug, or trust the system.

AI inbox gatekeepers will turn email marketing into earned media

For years, marketers have treated email as ‘owned media’: you build a list, you control when to send, you own the access.

But that model is breaking down. As marketers and SDRs use AI to generate more messages faster (over 376 billion emails were sent daily in 2025, roughly half of them unwelcome), buyers are fighting back. Google’s Automatic Extraction may override preview text with AI-generated summaries of deals and offers; Yahoo Mail replaces subject lines entirely with AI summaries. Apple Mail bundles promotional emails from the same brand into a single group, showing only an AI-generated summary. And tools like Fyxer AI and Outlook Copilot go further, triaging incoming mail, drafting replies, and filtering out unwanted messages before a human ever sees them.

Buyers aren’t deploying these AI gatekeepers out of spite; they’re drowning in email volume, and taking back control.

Prediction: In 2026, email marketing will shift from “owned” media to earned media, as AI gatekeepers enforce what was always true: inbox attention is granted based on relevance, value, and trust.

Four implications for B2B marketers:

• Fewer emails, more value. The volume game is over. If you’re sending weekly emails that rehash the same positioning, you’re training AI filters to deprioritize you. Instead, email when you have something worth saying: a useful insight, relevant research, a genuinely helpful resource. When subscribers recognize that your emails actually help them, both they and their AI assistants will whitelist you.
  
• Send from real people. Plain-text emails from your CEO or a named sales rep will often outperform polished HTML marketing sends. They feel like human-to-human communication, which encourages replies and actual conversations. They’re also lighter on code, which means they load faster on mobile and avoid spam triggers that flag heavily formatted emails.
  
• Optimize for AI readers. An algorithm reads your email before any human does. Front-load your key point in the opening sentence; AI summarizers grab the beginning to determine relevance (busy executives skim the same way). Straightforward subject lines will survive AI rewriting better than clever hooks. Use Gmail and Yahoo’s promotional schema so your offers show up accurately in AI-generated previews. And make your sender name immediately recognizable; it might be the only identifier visible in a summarized view.
  
• Measure what matters. Open rates were already unreliable; AI-mediated inboxes make them useless. Focus on actions that indicate real engagement: click-throughs to specific content, replies to your emails, meetings booked, opportunities created.

Email isn’t going away. Nearly 4.6 billion people use it today, projected to hit 5 billion by 2028 (Sopro.io). But the days of batch-and-blast are finished, while quality email will survive. In fact, if you’re sending genuinely helpful emails, AI filtering may work in your favor by removing the noise that makes buyers otherwise tune out entirely.

Taste, trust, and accountability will become the antidote to AI slop

AI has made content creation nearly free. The result is feeds filled with ‘AI slop’: content that looks polished but offers nothing, even when superficially personalized. 

Prediction: In 2026, buyers will use the source as a proxy to decide whether something is worth their time, whitelisting voices they trust and ignoring the rest.

Forrester’s “Predictions 2026: B2B Marketing, Sales, And Product” confirms this shift is already underway: “Trust has fragmented, with B2B customers relying more on personal networks and curated sources than institutions or broad brand promises.” 

The source is becoming as valuable as the substance. Three attributes will separate signal from noise:

Taste — knowing what’s worthwhile. It’s the ability to make discerning judgments about quality and value, a guide that helps determine what’s meaningful. Brandon Short describes it as “a blend of technical capability, editorial sensibility, design instinct, and storytelling.” As David Brier says, it’s the shift from “artificial intelligence to emotional intelligence”.

Trust — the relationship you’ve built with your audience over time. People trust people, not logos, which is why human emails outperform HTML and why individual LinkedIn posts beat your corporate account. It’s also why people subscribe to (and sometimes pay for) Substack newsletters from individuals while ignoring most vendor communications. This is why investing in founder brands and executive influence has become strategic, not optional — the founding creator is becoming as important as the founding engineer. It’s also why influencer marketing (long a B2C staple) is gaining traction in B2B: Forrester predicts 75% of enterprise B2B companies will increase budgets for influencer relations in 2026.

Accountability — the willingness to stake your reputation on what you share. Think of a court reporter: AI could transcribe a courtroom perfectly, but we don’t call it the official record until a human puts their name on it. The same logic applies to any professional using AI for research or content. The output is easy; the value is in standing behind it.

What this means for marketing leaders:

• Lean into communities and partnerships. No AI summary can replace a trusted peer saying “this solution worked for us.” Ecosystem-led growth and B2B influencer programs will continue to see increased investment.
• Prioritize human connection. When digital content is trivially easy to produce and filter, in-person moments become the differentiator: executive dinners, on-site events, conferences.
• Signal craftsmanship. Content that clearly required significant human effort and expertise signals substance over slop. Handwritten notes, proprietary data, deep technical analysis. When anything can be generated quickly, “good enough” isn’t enough, and evidence of genuine work becomes a trust signal.

Forrester framed it well: in 2026, B2B marketers must shift from persuasion to proof. The scarce resources are human judgment, relationships, and the willingness to be accountable for what you share.

Public intent signals will commoditize; proprietary signals generate “alpha”

When every team has access to the same intent data, that data stops being an advantage. Job changes, funding rounds, website visits, G2 activity — these signals are now available to anyone willing to pay for them. And AI-driven enrichment tools have made it cheaper than ever to process signals at scale, causing outbound volume to explode. 

Prediction: Generic signals based on public data will become commoditized within the next 24 months. Competitive advantage will shift to proprietary signals, signal combinations, and timing precision.

Investors call this ‘alpha’, meaning the extra return that comes from information others don’t have. Once everyone has the same information, the alpha disappears. (Hat tip to Brendan Short for this analogy.)

The new alpha comes from three places:

• First-party signals. Interactions with your own sales and marketing activity are proprietary by definition. Demo requests, content engagement, free trial signups, interactive product tours, inbound inquiries — these signals tell you something no competitor can buy. Instead of treating inbound interest as a ‘hot MQL’, treat it as one strong timing signal among many.
  
• Signal combination matters more than any single signal. A job posting alone is table stakes. But a job posting plus engagement with a competitor comparison page plus a spike in web visits from multiple stakeholders from the same account? That combination tells a story no individual signal can. Public signals may be commoditized individually, but how you mix them with your private signals creates an edge.
  
• Timing precision beats content personalization. As I wrote earlier, real personalization means the right action at the right time, not AI-generated emails mentioning someone’s LinkedIn post. The same principle applies to signals. The value isn’t just knowing who to contact or even what to say, but when. Niche signals that indicate the right moment have staying power that generic signals don’t.
  

Where martech is heading beyond 2026

The predictions above describe where martech is going: marketing to agents, composable stacks, AI-powered journey orchestration, and reasoning AI replacing rules. But what does it look like when it comes together?

Prediction: The future of marketing platforms is signal-based orchestration: ingesting signals from across the data ecosystem; deciding the optimal sequence of actions for each account, person and agent; and orchestrating execution across channels.

The three layers

Data. As briefly discussed in the composability prediction, the foundation is data — but what drives value is turning that data into usable signals across multiple sources: CRM, product usage, online behavior, third-party data, and unstructured sources (web scraping, LinkedIn profiles, call recording, etc.). Where that data physically lives is less important than making it clean and actionable for marketing. Maybe your cloud data warehouse handles this perfectly. More likely, your marketing platform will need to connect to multiple sources and use reasoning AI to turn raw data into signals the Decisioning layer can act on.

Decisioning. This is the brain that computes optimal multi-step sequences of actions, given all the signals plus the Strategies set by humans (what KPIs to optimize for each segment, plus constraints like frequency and budget). Multiple AI agents will work together: deciding which offer to deploy, how to personalize content, which channel to use, when to send, how to manage frequency, what actions need human review and what can proceed automatically.

Delivery. Channels will become execution endpoints, accessed via API. But they won’t be dumb pipes. Many will retain their own intelligence for channel-specific optimization: a DSP manages bid strategies, an email platform handles deliverability and send-time optimization, an ad network manages frequency within its ecosystem. The Decisioning layer orchestrates across channels; each channel may still optimize within its domain.

One requirement across all three layers: much of B2B buying happens while buyers are researching anonymously, so the architecture must handle account-level signals and actions when appropriate.

Note: the large cloud data warehouse vendors (Snowflake and Databricks) won’t be content just owning the Data layer for reporting and analytics. They’re pushing up the stack toward Decisioning (Orchestration) and other marketing use cases. This sets up an inherent tension: if every vendor wants to own journey orchestration, which wins in the long run? The battle will be confusing and messy as this sorts itself out, but it should ultimately produce better, more capable platforms for everyone.

What this means for 2026

The full vision is years away, but you can start building toward it now: invest in your signal foundation, capture operational context so AI can learn how your business works, curate a library of quality offers with clear tagging, and honestly assess whether your current platform can evolve with you or will hold you back.

Rapid-fire mini-predictions

1. More executives will finally realize that pipeline problems are actually positioning problems — and that no amount of tactics can compensate for weak messaging and unclear differentiation.
   
2. Attribution modeling will further decline as companies accept the truth: buying journeys are too non-linear and complex to meaningfully assign credit to specific touches.
   
3. AI experimentation budgets are driving a boom in AI pilots, despite the broader trend toward stack consolidation. But expect high churn since add-on tools are inherently less sticky than core platforms.
   
4. AI agents will take over many customer support interactions, making human contact a premium offering. As Angelo Robles puts it: AI-free becomes the new GMO-free.
   
5. We will see complete photoshoots and videos generated by AI. Humans will still need to guide creative direction, but the actual production will be automated without models, actors, etc.
   

AI-driven uncertainty will intensify in 2026; preparation is the only answer

Let’s finish the predictions by looking beyond GTM. Last year, I was optimistic; I believed AI would push us to embrace our uniquely human capabilities — creativity, emotional intelligence, strategic thinking, and genuine expertise. I still believe that potential exists.

But I also find myself more worried about the world than I’ve been in my entire career.

The job of the entrepreneur is to have a vision of the future and build toward it. That job has never felt harder. How do you plan when the goalposts keep shifting? When last quarter’s assumptions feel instantly obsolete? When the future is more uncertain than ever?

In fact, the World Uncertainty Index has spiked to levels that dwarf anything in recent history, including both the 2008 financial crisis and the early days of COVID. This uncertainty goes beyond AI: economic volatility, geopolitical instability, and shifts in customer behavior all contribute. But AI deepfakes eroding trust, growing loneliness as people turn to AI instead of each other, and especially the job displacement I’ll discuss below all make it worse.

Prediction: AI-driven disruption and global uncertainty will intensify through 2026. As Ethan Mollick argues, even if AI development stopped tomorrow, we’d still face “massive and rolling disruption across society and the economy for the next ten years” as organizations figure out how to harness what AI can do.

But AI development won’t stop. Progress in 2026 will come less from throwing raw compute at ever-larger models, and more from specialized architectures, better workflow integration, improved long-term memory, and increasingly capable autonomous agents. As they say, today’s AI is the worst AI you will ever use.

Job displacement

Will AI eliminate jobs in 2026? Probably not yet, at least for the complex jobs like those of the people who will read these predictions. Many jobs, including GTM, involve a mix of tasks: strategy, creativity, relationship building, analysis, administration. AI handling some of these will shift what we do; it doesn’t eliminate the position.

But that’s not true for entry-level roles. A Stanford Digital Economy Lab study showed a 16% to 20% decline in employment for AI-exposed positions like software development, marketing, and customer service, with the impact concentrated among people early in their careers. Pave data shows SDR positions dropped from 1.98% of the workforce in January 2023 to 1.45% by August 2025. At large tech companies, employees aged 21-25 fell from 15% to just 6.8%.

Angelo Robles calls this “The Silent Freeze”: companies maintain productivity without backfilling junior roles. The issue isn’t layoffs; it’s the destruction of the on-ramp. If juniors aren’t hired, they don’t become seniors. We risk creating a barbell economy: massive demand for AI-proof trades (plumbers, electricians) at one end, high-level strategists at the other, and a hollowed-out middle class.

And this won’t stay confined to entry-level work. Chris Penn points to the Remote Labor Index, which measures whether AI agents can accomplish complex commissioned projects at commercially acceptable quality. Current models score around 2%, but if models improve the way they did on other benchmarks last year, that becomes 20% in 2026. At that point, we’re talking real displacement moving up the skill ladder. And what happens when robotics has its ChatGPT moment and becomes commercially viable? When self-driving cars become a reality? Suddenly whole swaths of the economy face additional disruption.

The societal stakes are serious. As Penn writes, historically, “when enough people have been displaced from work in a very short period, that’s when things like pitchforks, torches, and guillotines tend to come out.” Combined with the broader uncertainty already gripping the world, this keeps me up at night.

Ideally, governments and tech companies would step in to help. Not by stopping or over-regulating AI, but with policies to mitigate job displacement. We’ve done this before: labor protections following the industrial revolution, the GI Bill after World War II, trade adjustment assistance during offshoring waves. 

We can’t freeze in the face of uncertainty

I don’t have a simple playbook for what comes next, but standing still isn’t an option either. Here’s what I believe leaders can do.

Build resilience through efficiency. When you don’t know what will happen, efficiency creates the cushion that lets you absorb shocks without breaking. Organizations running lean can redirect resources quickly; those already stretched thin have no room to maneuver. Drive productivity improvements now and bank the savings.

Focus on enabling people. Teach teams to use AI, to explore possibilities and challenge assumptions rather than accepting the first generic answer. As Trust Insights emphasizes, effective AI use demands subject matter expertise, data fluency, and intimate customer knowledge; vague prompts produce undifferentiated results.

Use AI itself for upskilling. AI is remarkably good at coaching, providing personalized, on-demand skill development that scales far beyond any training department. Use this to accelerate the growth of your existing team.

Rethink team structure and management roles. The era of hyper-specialized siloed roles is ending. Marketing needs generalists who think strategically, create compelling work, and use AI tools fluidly across data, content, and execution. Org design becomes more about adaptability than headcount; managing AI agents increasingly feels like HR-style governance.

Frontline management changes too. As AI absorbs management tasks like reporting, coaching, and routine inspection, managers shift to customer-facing leaders. The goal: deeper customer insight, cleaner execution, and more time where human judgment matters most. 

Conclusion

As analyst Nicolas de Kouchkovsky puts it, “2026 won’t be predicted as much as navigated. The ground is shifting faster than teams can make confident forecasts.” 

I don’t know exactly what is coming; that’s the nature of uncertainty. My predictions in this article are themselves uncertain, of course. I’ll grade them next December, just as I’ve done for the past three years, and some will prove wrong. But they represent my best assessment of where things are heading, and they’re shaping how I’m navigating the year ahead. I hope they help you navigate yours.

Happy New Year!

Five years ago, Jason Baldwin and I published Martech 2030 (ungated PDF), a report predicting five major trends that would shape marketing and martech through what we christened The Decade of the Augmented Marketer.

We’re now at halftime at the start of 2026.

How are those predictions holding up? I asked ChatGPT, Claude, and Gemini to grade them. Their assessments were pretty favorable — especially for our vision of “harmonizing humans & machines.” (It’s worth a chuckle that Claude and Gemini gave us A+’s on that one. Very harmonious.)

But you should be the real judge. How do these line up with your lived reality?

Trend #1: “No Code” Citizen Creators

Even in the previous decade, I was already bullish about “no code” platforms empowering marketers to self-service implementation of their digital ideas. Airtable, Webflow, Zapier were my go-to examples for databases, websites, and workflow automations respectively. Gartner called these users citizen creators. I’ve called them “marketing makers,” builders in the go-to-market realm. And I expected that we’d see a lot more of them — with a lot more power at their fingertips.

In all modesty, we nailed this one. But largely due to a development that we didn’t see coming: the generative AI explosion that started in late 2022.

No code used to be mostly drag-and-drop. With LLMs, suddenly you could simply describe what you wanted, and the AI would create it. At first, it was mostly for creating text and six-fingered images (paging Inigo Montoya). But over the past two years, you could talk your way into building apps, agents, videos, infographics, workflows, campaigns, data analyses, PowerPoint decks, etc. It’s still a jagged frontier, but the output for many of these is getting quite good.

As Andrej Karpathy said, “The hottest new programming language is English.”

The vibe coding phenomenon of 2025 — a term coined by Karpathy — pushed this into hyperdrive. Bolt, Lovable, Replit, and Vercel made it insanely easy for non-engineers to create apps and agents. Granted, this worked best for relatively basic use cases. But those were exactly the kind of underserved opportunities that Clay Christensen had observed as the foundation of disruptive innovation. See my post on the Lemkin Scale of Vibe Coding for a deeper dive on this.

In many ways, the major AI assistants such as ChatGPT, Claude, and Gemini are über no code platforms, letting you create an ever-wider range of digital assets on demand. Teams who leaned into these capabilities have reaped the benefits of greater speed, bandwidth, creativity, and learning of decentralized self-service.

That said, governance and guardrails for citizen-created assets remain immature. Many organizations are still figuring out how to balance empowerment with brand consistency and compliance. We’ll come back to this subject with Trend #4 about Big Ops…

Trend #2: Platforms, Networks & Marketplaces

If prior decades were about firms optimizing linear chains — value chains, supply chains, distribution chains — we believed the structure and strategy of businesses would evolve to increasingly be driven by interconnected graphs and ecosystems via platforms, networks, and marketplaces.

In martech, platform ecosystems have now fully supplanted the old suite vs. best-of-breed dichotomy. (I was delighted to contribute to this shift as VP platform ecosystem of HubSpot, a role I left this past September after 8 years of building the company’s ISV program from the ground up.)

Two of the hottest platforms today are Databricks and Snowflake, which serve as unviersal data layers that integrate across heterogeneous tech stacks. Probably the most intriguing platforms headed into 2026 though are the major AI engines, with MCP connectors and the emerging concept of ChatGPT Apps.

Networks were already ubiquitous, but have grown even more so. Digital marketplaces have proliferated. I would say the most noteworthy ones in martech are the hyperscalers: AWS Marketplace, Google Cloud Marketplace, and Microsoft Marketplace now transact an estimated $45 billion in annual enterprise software sales. Adobe, HubSpot, and Salesforce now all offer products through this channel.

One seemingly paradoxical balancing act of platforms, networks, and marketplaces: it is through the centralized control of standards and governance, when paired with technical and philosophical openness, that these ecosystem mechnisms drive decentralized growth and innovation. You centralize to decentralize. It’s a bit of a Zen koan for the digital age.

That’s partly why this is only an A-/B+ so far. Ecosystem dynamics are still new for many companies, who prefer the greater control they have — or at least the illusion of control — with more direct, linear go-to-market strategies. But those are old mental models from a previous era. I’ll double down on my bet that ecosystems and the platforms, networks, and marketplaces that animate them will underpin marketing by the end of this decade.

Trend #3: The Great App Explosion

When we wrote the Martech 2030 report, the 2020 martech landscape had 8,000 solutions. As part of an annual tradition, most people assumed that was “peak martech.” But here we are in 2025 with nearly twice as many: 15,384 commercial solutions on the map.

Why hasn’t this consolidated? Actually, it continually consolidates through competition and M&A. But because there are effectively no barriers to entry in software — a trend that AI has dialed up to 11 — new products continue to pour in at a faster rate than they drain out.

Ironically, it is the consolidation of hyperscaler cloud infrastructure and major app platforms that spurred the blossoming of thousands of specialist apps built on those foundations. (Centralize to decentralize.) These apps make up the long tail (the loooooooong, long tail) of the martech landscape: vertical market apps, regional apps, ecosystem apps, service provider apps, and an unending tide of startups aspiring to disrupt the incumbents.

But we predicted that The Great App Explosion would be even bigger. Not thousands of apps. Billions of apps. The vast majority custom built by organizations, tailored to their own operations and customer experiences. While we expect the long tail of commercial martech will continue to grow, it will be dwarfed by the hypertail of custom martech. (For more about the hypertail, download our State of Martech 2025 and Martech for 2026 reports from last spring and this past December.)

AI has delivered The Great App Explosion, and the hypertail is already thriving today. Five predictions we made within this trend are also materializing, including service-as-software (takeaway #3) and the rise of intelligent/interactive marketing assets (takeaway #4):

Trend #4: From Big Data to Big Ops

We anticipated that the 2020’s would be a decade of Big Ops. Analogous to wrangling the scale and complexity of Big Data, the mission of Big Ops would be to tackle and tame the scale and complexity of hundreds to thousands of parallel apps, agents, and automations operating on that sea of data within a company’s operations.

We correctly called the problem — the chaos of uncoordinated “No Code” Citizen Creators and The Great App Explosion — earning ourselves a solid B/B+. Solutions, both technical and organizational, however, are still emerging. Almost every major platform in martech is focused on this problem/opportunity right now, framed as “orchestration” and “decisioning.”

This is going to be the platform battle for the rest of this decade. Grab your popcorn.

Data is a critical asset in Big Ops. But its value is only unlocked through its distillation and activation, which we described as data intelligence and data relfexes. One of my favorite metaphors from our report: data is not the new oil, it is the new oil paint. Big Ops is how we go from pigments on a palette to masterpieces on a canvas.

It’s the entanglement of Big Data and Big Ops that positions Databricks and Snowflake — as well as the hyperscalers AWS, Google, and Microsoft — as serious contenders in the Orchestration Wars.

It will be a frenzy of coopetition between them and the public martechs (e.g., Adobe, Braze, HubSpot, Salesforce, Zeta), evolving CDPs (e.g., Hightouch, Tealium, Syncari, Treasure Data, Twilio), agent/automation platforms (e.g., CrewAI, n8n, Make, Workato, Zapier), a wave of next-gen martech startups (e.g., GrowthLoop, iCustomer), and even the AI giants Anthropic and OpenAI.

Going to need more popcorn.

Trend #5: Harmonizing Humans + Machines

“Greater automation and AI assistance will give marketers more time and new tools to focus on customers, creativity & innovation.” If we made that prediction today, the reaction would be “no duh.” But five years ago, this was prescient. The graph we sketched below is pretty much exactly how this is playing out.

We noted that automation and human engagement aren’t necessarily mutually exclusive, and when properly synthesized together can deliver magical experiences:

We also anticipated the rise of buyer-side AI, which emerged as a market force in 2025 with Agents of Customers. What we envisioned in 2020 as a shift from SEO to BBO (Buyer Bot Optimization) is the phenomenon we’ve come to know as SEO to AEO (AI Engine Optimization). What we predicted as Bot Commerce is now called Agentic Commerce.

The labels we’ve landed on are cosmetically different, but the AI concepts we defined five years ago have crossed from theory into reality. I think we earned our A+’s here, no?

Conclusion

What do you think? Did the ChatGPT, Claude, and Gemini grade us fairly?

Amara’s Law states that we tend to overestimate the short-term impact of technologies but underestimate their long-term effects. The goal we had with our Martech 2030 paper was to think about that long-term arc. While these five trends are still evolving, it’s remarkable how far they’ve come in these past few years. The pace is quickening.

For better and worse, I don’t think it will slow down in 2026.

Let’s revisit this paper one more time in 2030 for final grades.

It’s here! Our brand-new, 127-page Martech for 2026 report.

No form to fill. No “leave a comment on LinkedIn” arm twisting. You can simply download the full PDF right here.

These are the 7 major subjects we cover to prepare you for 2026:

1. The hype-free reality of SaaS & AI in martech
2. 3 types of AI agents in marketing (only one is disruptive)
3. Context engineering in marketing & data sources in use
4. A spectrum of deterministic & non-deterministic automation
5. Practical application of “vibe coding” in marketing
6. “The Factory” vs. “The Laboratory” in martech management
7. Value engineering > GTM engineering in Marketing Ops 3.0

It includes all the latest data from our AI & Data in Marketing survey conducted a couple of months ago. No techno-jargon. Just plain language explanations and helpful visual illustrations.

For a report of this depth and rigor, we could easily charge $1,000 or more. Instead, we worked with 7 sponsors to support the months of work we spent on this to be able to share it with you at no cost.

Our sponsors had no input or review of our editorial content — they just saw the report for the first time yesterday. Instead, we conducted a series of interviews with leaders from each of them to dig into their perspective on these subjects, which we included as separate chapters in the report:

GrowthLoop — Mind the Gap: Why AI Success Starts with Data
Hightouch — AI Agents and the Future of Marketing Workflows
Intuit Mailchimp — AI and the Mid-Market Marketing Revolution
MetaRouter — AI and the First-Mile in Digital Experiences
Progress — The Human Dimensions of AI in Martech
SAS — With AI, More Is Not Better — Better Is Better
Treasure Data — The Future of AI and CDPs in Marketing

Frans Riemersma and I learned a ton working on this. We hope it’s genuinely useful for you, and we’d love to hear any feedback you have for us.

It’s almost here! We’re launching our in-depth Martech for 2026 report on December 2, and you can be among the first to see it. Simply sign up for our Martech for 2026 webinar, and you’ll get the full report the instant it’s released — along with our scintillating repartee of key findings that you can catch live or on-demand.

To share a sneak preview with you, we’ll affirm: AI agents in marketing are real.

90.3% of the participants in our AI & Data in Marketing study reported they’re using AI agents somewhere in their martech stack, even if it’s only for experimentation or in limited production use cases.

Granted, our participants skew more tech-savvy than the average marketing team. But they’re running martech and marketing operations at some of the top B2C and B2B brands in the world. They take their responsibilities seriously. They’re not starry-eyed or reckless. They’re pragmatic and competing at the top of their game.

So what are the AI agents in marketing’s orbit?

To us, the defining characteristics of AI agents in marketing aren’t how big they are or how many tools they use. It’s the people in their lives. Who do they work for? Who do they interact with? Who do they benefit?

The vast majority of AI agents in marketing work for marketers. Hundreds of new agentic tools — plus hundreds of agents added to existing martech platforms — that can help marketers brainstorm, produce, distribute, and analyze campaigns and programs.

The mission of these agents is to improve marketing. In many cases, they make it more efficient. (By the way, don’t you just love the “do more with less” cliché that masquerades as AI strategy with some executives? We thought so.) In the best cases, they actually make marketing more effective too.

But almost all of these improvements are incremental, even if they’re significant.

The real disruption to marketing is happening on the other side of the equation: AI agents that buyers control. The most obvious example of this is the displacement of classic Google search — and the SEO playbooks that marketers have spent the past 25 years honing to win that channel — by AI assistants such as ChatGPT, Claude, Perplexity, and Google’s own Gemini platform.

Buyer-side AI agents are true disruptors because they are fundamentally changing the way in which buyers find, evaluate, and engage with B2C brands and B2B vendors. In this Great Rewiring of the Buyer’s Journey, marketers are losing the visibility and control we’ve become accustomed to.

A whole category of AEO products has sprung up to help address this challenge, growing rapidly. But they’re early in their evolution, and the environment of buyer-side AI agents is advancing at an even faster rate. It’s like chasing a bullet train with a bicycle.

And in our Martech for 2026 research, while we found that the majority of marketing teams (63%) recognize this shift in buyer search channels and behaviors and are intentionally adapting their content strategies accordingly, very few (14%) have instrumented this yet. We expect this will get prioritized in 2026.

We believe buyer-side AI agent disruption is only going to accelerate over the next 6-12 months. Already, AI assistants are morphing from informational answer engines to agentic action engines that can engage with agents or humans on the seller’s side in the service of a customer’s goals.

OpenAI launched their Instant Checkout feature a couple of months ago, enabling users to purchase products from Etsy and Shopify stores directly within ChatGPT. A deal to do the same with Walmart was announced a few weeks later. In parallel with ChatGPT Apps, which embed 3rd-party services directly into ChatGPT’s conversational interface — initial apps include Booking.com, Canva, Coursera, Figma, Spotify, and Zillow — users can increasingly bypass a vendor’s own web experience with their AI assistant.

That’s a very different buyer’s journey!

Google has launched similar transactional functionality with their agentic checkout feature, which can patiently track prices for a desired purchase until it falls within your specified budget and then automatically buy it for you with Google Pay.

One of the new features with Google’s Gemini AI agent is the option to have it call local stores to check for pricing and availability of products you’re interested in. Yes, Google will make voice calls to multiple stores, ask the humans they connect with questions on your behalf, and then report back their answers and offers to you over email.

You have to appreciate the irony in this turning of the tables. For the past year, a number of B2B sellers have been giddy about deploying outbound AI SDRs to tirelessly reach out to prospects and only alert one of their human sales reps when the AI snags a fish on the line. No concern about the fact that human prospects on the receiving end had to wade through more “personalized” (but actually impersonal) incoming emails, texts, and phone calls. The goal was efficiency for reps, not for customers.

Ah, but now, the buyer can sit back at the spa while their AI agent calls human salespeople on their behalf. New efficiency for the customer. Not so much for the salespeople who now face an increasing volume of incoming calls from AI agents because, hey, no sweat for the buyer. Now, they may choose to hang up. But then they’re foregoing a possible sale if one of their competitors is willing to take the call.

Frankly, having AI agents harass engage humans in either direction isn’t particularly efficient. We believe this will be a very brief intermediate state that will hasten the adoption of agent-to-agent interactions.

That’s going to be a whole new game for marketers. How do you market to and serve AI agents as intelligent intermediaries working on behalf of your target audience?

We’ll talk through the first pragmatic steps of that journey in our Martech for 2026 report and webinar. I’ve only touched on a tiny slice of our research here. Come join us on December 2 to step back and get the full picture of what the year ahead has in store for us.

Sign up here for the report and the webinar (live or on-demand).

(A huge thank you to our sponsors GrowthLoop, Hightouch, Intuit Mailchimp, MetaRouter, Progress, SAS, and Treasure Data, whose support for our research enables us to provide this 100+ page report at no cost to you.)

Every December for the past few years, Frans Riemersma and I have published a deeply researched report on what’s actually happening in martech as the New Year approaches.

These aren’t the crazy fever-dream “predictions” that you’re used to reading during this eye-rolling, end-of-year prophecy season. It’s a pragmatic assessment of what’s really happening inside marketing tech stacks and the martech industry as a whole. Existing, real-world patterns that can be reasonably extrapolated for the next 6-12 months to get you prepped for the year ahead.

Our latest edition — Martech for 2026 — is almost here. We’ll be releasing it on Tuesday, December 2. You can be one of the first to get a copy by signing up for our accompanying free online event. Frans and I will give a no-holds-barred presentation to walk through our latest and most surprising findings. If you register in advance, you can catch us live or on-demand anytime that month.

We also promise to make it fun, because a sense of humor makes it all more digestible.

One of the major topics we will cover in considerable depth — surprise, surprise — are the plethora of new AI agents on the scene and their relationship within existing martech stacks and marketing operations practices.

We’ll analyze three different domains of AI agents in marketing’s orbit:

1. Agents for Marketers — AI that marketers are using behind-the-scenes to design, produce, and analyze marketing campaigns, programs, and experiences.
2. Agents for Customers — marketer-controlled AI that interacts directly with customers, such as customer service agents, AI SDRs, and website concierges.
3. Agents of Customers — buyer-side AI that customers use to research, evaluate, and even negotiate and purchase from brands and vendors, outside of marketing’s control. This is where real disruption is happening! AEO is just the tip of the iceberg.

We’ll share the complete results from our AI & Data in Marketing survey to reveal which of these agents are scaling in production and which ones are still emerging from the science fair. We’ll dig into the art of “context engineering” and bring the stats about the full spectrum of data sources being connected to these agents.

We’ll also present real-world insights into the evolving relationship between these new AI agents and existing SaaS platforms in marketers’ tech stacks. Is AI augmenting those apps? Replacing them? Addressing entirely new use cases? We’ll take you through all of it.

And much, much more.

You get it all — our 100+ page report and our jam-packed, hour-long presentation narrating the best of it — with a quick, free registration.

We’re able to provide this epic report and its acccompanying online seminar for free thanks to the support of our sponsors: GrowthLoop, Hightouch, Intuit Mailchimp, MetaRouter, Progress, SAS, and Treasure Data.

BONUS: After Frans and I finish our keynote presentation, we will interview senior product and marketing leaders from each of these companies to get their perspective. No sales pitches, just candid discussions about what they’re seeing with AI with their customers and inside their own organizations.

Don’t miss this opportunity to get ahead of the curve for 2026 before your holiday break! Register now for Martech for 2026 to be at the front of the line for the new report and catch any or all of our sessions live or on-demand for the month of December.

I hope we see you there!

The lights are on but nobody’s home. The engine’s running but no one’s driving. The wheel is spinning but the hamster is dead. Pick your preferred idiom for unattended automations run amok. (“The Wi-Fi’s on, but there’s no Internet.”)

Over the past weeks, I encountered AI agents and automation at two separate martech companies — as a customer of one and a pseudo-prospect for the other — that leaned toward the “annoyingly efficient” upper-left quadrant of the 2×2 below. Efficient for the company. But for the customer? Not so much.

I’ll share what happened, as examples of how these things can slip off the rails, and then offer a few ideas for avoiding such outcomes. I won’t name names, because I generally admire both companies and think these kind of issues are not uncommon.

That said, yeah, it’s ironic that these are martech companies.

Customer service on the route to nowhere

A product I subscribe to has been emailing me helpful onboarding tips for weeks. It’s been a good cadence, not too frequent, and the content has been genuinely useful. They come from an email address that we’ll call friends@acme.com (not the actual company or email address).

However, last week I had a billing question. I searched for how to contact someone about it, and everything pointed me to emailing friends@acme.com. Simple and easy, right? So I emailed them my question.

No reply. Not even an automated acknowledgement of receipt.

Not a big deal. I wait a day, and then email again to check back.

Again, no reply.

A couple of days later though, friends@acme.com sends me another cheerful onboarding message. Really?

I reply once more asking about my billing issue. I volunteer that maybe I need to contact a different email address about this? Again, no reply.

I go back to hunting around their app to see if somehow this was my fault, that I missed a “for billing questions, contact billing@acme.com” instruction. I search FAQs, account screens, help docs. Everything just points to friends@acme.com.

I abandon the effort for the time being, as I had other things to do.

While it’s impossible to debug this from where I sit, it seems like an automated routing problem. Multiple teams sharing the same inbox with some automation rules to direct incoming messages to the right place. But it’s broken. Maybe it’s a bug in the routing logic. Maybe the destination I got routed to is unmanned. Maybe an LLM is trying to interpret my message but it’s erroring out. Perhaps a RAG look up of my account is failing?

Automations can be fragile, especially complex ones. As a martech person, I appreciate that. I empathize with this certainly unintentional fail. But better safeguards could and should have been implemented.

Funny enough, as I’m writing this, I just received yet another onboarding email from friends@acme.com, clearly still oblivious to my prior attempts to engage with them. I laugh out loud, causing the person next to me on the plane to give me a sideways glance.

To err is human, but to really mix things up use AI

Meanwhile, in another part of the martech galaxy, I downloaded a new report that a vendor was offering on a subject of great interest to me.

Now, in fairness to the vendor, I am probably an outlier responding to this classic demand-gen tactic. I don’t think of myself as a prospect for their product, at least not right now. I’m just interested in the subject matter.

Within a couple of minutes of filling out the form, I get a personalized email from a rep.

You want to talk hyperpersonalization? This message nailed it. The opening paragraph was crafted about the nature of my work on chiefmartec. It raised a concern about AI that was very relevant to me. It asked a genuine-sounding question about what sparked my interest in the report. It was a very human sounding message. No simple Hello, $FIRST-NAME template. This was production-quality generative AI personalization.

The only AI tell? The near immediate arrival. No human could have researched me and crafted that message that quickly. But, hey, that’s not a bad thing. A fast reply. A well-personalized message. I admire the fact that it was AI-powered.

So I reply, complimenting them on their response time — my wink-wink that I know how they did that — and letting them know what my interest was and why I’m probably not a real prospect for them. I assume that will make it to the real rep.

They didn’t reply back.

That’s fine. I had intentionally disqualified myself. But it felt a little odd. Their original email was very conversational, and I had engaged with a remark that I would have expected to elicit a response back. In exchanges I’ve had with reps at other martech companies under similar circumstances, they almost always replied. Those threads sometimes led to other beneficial connections and outcomes over time. (The funnel is perennially oversimplified.)

But no worries. They’d moved on, and I moved on.

Until a couple hours later when I got another email from the same rep — but formatted very differently. It opened with old-school Hello, $FIRST-NAME personalization, referenced something unrelated to what I downloaded, and offered calendar options for scheduling a demo. Huh? It was a total non sequitur to the exchange we just had.

Now, being a martech nerd, I can call this: I got thrown into an automated sales sequence.

I shake my head. A week later I get another message from the rep, seemingly sent by their automation engine, as it’s formatted the same as that last one. This one makes a generous offer, but for something pretty irrelevant to me. It’s also Hello, $FIRST-NAME personalized — but now also with a $COMPANY-NAME personalization plugged into the first paragraph. Except the company name inserted is most definitely not my company.

Yikes.

Cross-contaminated data? Misconfigured automation sequences? An LLM hallucination inside an AI agent? I suspect there may have been compounding errors along this journey.

My guess as to what happened? An AI SDR agent crafted the first message. The rep never saw it. Nor my reply, which the AI SDR intercepted and didn’t know what to do. So it just enrolled me in a sequence. It added some fields to my record through a data enrichment service that were just flat-out wrong. The sequence had no “memory” of my exchange with the AI SDR. These were probably two different products with a weak integration at best.

Learning how to leverage hybrid human + AI teams

Humbling moments like these are good to keep things real.

For the record, I am in favor of leveraging AI and automation in martech. That’s arguably been the raison d’être of this blog for the past 17 years. But for nearly as long, I’ve seen Martec’s Law at play: technology changes quickly, but organizations change slowly. As a corollary: we almost always underestimate the people and process changes required to successfully leverage fast-changing technologies.

With all the vague-but-urgent, top-down edicts to adopt AI, teams sometimes feel the need to rush new capabilties into production. The move-fast-and-break-things ethos. For real AI transformation though, I recommend a go-slow-to-go-fast ethos instead.

That doesn’t mean the teams driving this transformation are lollygagging about. It’s a high-gear intensity for everyone in this period of disruption. But you can’t skip over major stages of the journey. You need to take it a step at a time to learn what the technology can and can’t do. But more importantly, you need to understand its dependencies and ripple effects. You need to figure out which legacy processes need to be updated — and which ones need to be entirely reinvented or gracefully put to pasture.

Move fast with pilot projects and experiments! But contain them as narrowly as possible at first. And absolutely keep a human-in-the-loop — even if it’s just post-hoc review of what happened — watching closely. Don’t scale prematurely. You want to make sure that your foundation is solid because that’s what makes it possible to reliably move faster in the next stage. Chasing ghosts and wrangling tech debt in production really gums up the works.

Here are a few concrete suggestions:

Always have a way to reach a real human. AI and automation can be efficient for customers as much as it is for the company — when it addresses the customer’s needs. But it’s never going to be 100%. The edge cases are infinite. And while not all of them will be valuble opportunities lost, some of them will be. If a customer wants to reach a human, you probably want them to as well. The lower-right corner in the 2×2 at the top of this article sums it up: go the extra mile for loyalty & love.

Run secret shopper exercises. Have people who take the journey as prospects or new customers, carefully recording their experience. Doing this with your own employees is a decent place to start. But ideally you want someone external with fresh eyes and no bias for internal politics to run the gauntlet. This is a little more difficult in B2B, given the data enrichment industrial complex that hones in on one’s identity, albeit with varying degrees of accuracy.

Regularly inspect every active AI agent or automation. Just as elevators and escalators require regular inespection, anything that moves your prospects and customers along in their journey should be periodically reviewed. Are the rules and data definitions being used still valid? Is the content and branding still relevant? What are the contexts in which it works — i.e., what could have happened before it, what might happen next? Timestamp and sign these inspections. Repeatable test harnesses and heuristics are good. But you also want the benefit of human judgment in that check-up.

If you use a rep’s name in communications, loop them in. I know, the whole point of AI agents and automations in such scenarios is to reduce the manual work for reps. But they can use inbox rules to reduce the noise until there’s a response from the recipient. Then it’s probably worth at least peeking at it, even if they let the automation carry on. It also makes it easy for them to occasionally review what is being sent in their name. After all, it’s their reptuation as well.

Distinguish between the “laboratory” and the “factory.” My frequent collaborator Frans Riemersma has made the insightful observation that there are two different missions with martech stacks. One is serving as the steady and reliable factory in production use cases to protect current revenue. The other is serving as a laboratory to run experiments with new or adapted use cases. Keep these two things distinct in how you manage them! Be wary of lab leaks.

I also think there’s a tremendous opportunity for existing martech platforms and new martech entrepreneurs to provide more governance and monitoring capabilities to address these issues. This is a big part of what good “orchestration” needs in our new AI age of marketing.

Building an automation or deploying an AI agent is getting easier by the day. Keeping them all aligned and working properly? Without good operational discipline though, that’s going to be like herding cats.

Ah, the Hype Cycle. No other curve is both so revered and reviled.

When it comes to generative AI’s Hype Cycle, especially its application in marketing, there are three things we need to acknowledge:

First, “gen AI” is not a single technology moving along the Hype Cycle. Is it at the peak of inflated expectations? The trough of disillusionment? The slope of enlightenment?

The answer is: all of the above. Different gen AI applications and use cases are at different stages along this roller coaster track.

Second, many of these generative AI applications and use cases are in their first generation. Our expectations for a great AI-powered chatbot today are modest compared to what we’ll expect of them in 2 years, 5 years, 10 years.

For instance, today, a good customer service chatbot can provide accurate answers to relatively straightforward questions. Future generations of these customer serivce AI agents may handle much more complex scenarios, may have more agency to take actions on a customer’s behalf, may autonomously engage in cross-sell and upsell motions, etc.

Each of those generations will likely follow its own Hype Cycle of inflated expectations, disillusionment, and enlightenment. We can be at the Plateau of Productivity for today’s chatbots, while facing the start of a fresh overhyped Hype Cycle for the next generation’s capabilities.

Visualizing multiple generations of a technology, each on their own Hype Cycle journey, lets us hold two opposing stages of the cycle in our mind at the same time (the test of a first-rate intelligence according to F. Scott Fitzgerald). AI agent chatbots can be simultaneously in the Plateau of Productivity and the Peak of Inflated Expectations.

Or more accurately, “a” plateau and “a” peak — each one of many over time.

Third, the speed at which AI applications are rocketing through the Hype Cycle is unprecedented. What used to be a decade-long journey from trigger to plateau now happens in a couple years. Before one generation is plateaued, we’re already off to the races with the next generation’s cycle.

I know you feel this acceleration in the pit of your stomach. Trust me, we all do.

But a new report released by SAS this week — Marketers and AI: Navigating New Depths — gives us some empirical data on the maturation rate of gen AI use cases in marketing.

I found this chart the most interesting, comparing adoption vs. planning of 10 different gen AI applications in marketing from 2024 to 2025 (n=300 for both years):

The 12-month acceleration in adoption of several of these is impressive:

• Trends analysis: +56.5%
• Chatbots/customer interactions: +44.2%
• Generating text/copy: +32.4%

Those three most accelerated use cases are also the three most adopted ones too, albeit in a slightly different order:

• Chatbots/customer interactions: 62%
• Generating text/copy: 45%
• Trends analysis: 36%

However, there were also a couple of use cases that reversed direction over this past year, showing a reduction in adoption:

• Generating video: -9.1%
• Customer journey mapping: -4.3%

These stats make it possible — albeit with a little hand waving — to map where on the Hype Cycle each of these use cases currently sits. The most adopted? They’re in a Plateau of Productivity. The ones losing adoption, especially after much fanfare a year ago? In a Trough of Disillusionment.

The image at the top of this post is my approximation of a Hype Cycle map for these use cases — at least in their first generation. Let me emphasize their first-generation-ness.

Also, to be crystal clear, Gartner didn’t make this map. They might be horrified by my use (abuse?) of their model. The placement of these gen AI use cases are my own estimates, based on the data from this new SAS report, as well as a collection of anecdotes I’ve heard from across the martech community.

Here’s a summary of my rationale for where I placed these use cases:

There’s plenty more data and insights in the full SAS report — including some fascinating stats on the early considerations of quantum computing in marketing. (In case you were worried we might run out of massively disruptive and transformative technologies for us to absorb in the years ahead.)

Isn’t it an exhilirating time to be in marketing and martech?

Are you feeling the vibes yet?

“Vibe” has become the latest AI buzzword. Even more than “agentic,” which is saying a lot. While both terms are widely open to interpretation — and therefore ripe to be appropriated by anyone selling anything — vibe stands out for sheer weirdness.

The term originated with a post by Andrej Karpathy, one of the original co-founders of OpenAI, who described vibe coding as a new way of rapidly building software by simply conversing with an LLM-powered coding assistant. “I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.”

I think of this as the next generation of no-code/low-code development, now turbocharged with AI. Which, hey, is awesome. “Power to the people” as they said in the 60’s.

I’m sure the “it mostly works” caveat is triggering to some. Critics of no-code have long argued that people who don’t know how to code have no business building their own apps. Taking a “just work around it or ask for random changes until it goes away” approach to fixing bugs would horrify them.

Andrej, however, is a world-class computer scientist who certainly knows how to code. Yet here he is advocating to give in to the vibes, embrace exponentials, and forget that the code even exists.

(As an aside: is “vibe” language a bizarre mashup of 1960’s counterculture with 2020’s techno-optimism, or what? I’m not saying that recreational drugs are involved. But if there were gummies other than Haribo next to the keyboard, I wouldn’t be shocked either.)

In recent months, vibe coding has become all the rage.

Jason Lemkin of SaaStr recently got in on the vibes, diving into a 100+ hours of vibe coding an app with Replit as a non-engineer — albeit with mixed results. He shared his joys and frustrations as he went along in a series of LinkedIn posts.

Afterwards, he wrote a great post about what kinds of apps make sense to be vibe coded by non-engineers, from safe “green light” apps such as internal dashboards or personal workflows to here-be-dragons “red light” apps, such as trying to rebuild Salesforce from the ground up. (I’m not sure if even Salesforce would attempt to rebuild Salesforce from the ground up.)

I converted this into a visual continuum that I call the Lemkin Scale of Vibe Coding, ranging from 1 (viable vibing) to 10 (unviable vibing):

Don’t read too much precision in this scale. It’s just a loose way to approximate the level of challenge a non-engineer would face in trying to vibe code a particular app. Comparable to when the doctor asks how much pain you’re feeling, on a scale of 1-10. (“Depends, doc. Ask me again after I see how much my insurance actually covers.”)

Of course, the examples along this spectrum are only a snapshot of what’s easy or hard today. As these AI tools continue to improve at an impressive clip, we can expect more advanced app building to “shift left” to the green side of the scale over time. What’s a “7” today might be a “3” in a year from now.

“Vibe Marketing” was inevitable, but what is it?

With vibe coding sweeping social media and garnering so much attention and love — the startup to achieve the fastest growth to $100 million ARR, in just 8 months since November 2024, is a vibe coding platform literally called Lovable — it didn’t take long for marketers to jump on the bandwagon with “vibe marketing.”

I think Greg Isenberg gets credit as the first person to coin the phrase.

But what exactly is vibe marketing? Or what should it be?

Greg framed vibe marketing as a combination of vibe coding with AI agents and workflows that empowers individual marketers to achieve things that previously would have required a whole team of specialists. They can move faster and ship more ideas at a fraction of the cost. They can easily build or customize tools and systems tailored to their specific needs. Vibe marketers are übermarketers. (Even if they don’t work at Uber.)

(As an aside, this is kind of a cliché marketing move: if “vibe” is cool, and “agentic” is cool, then something that is both vibe and agentic has to be even more cool, right?)

If vibe coding is the next generation of no-code/low-code, this definition of vibe marketing seems like the next generation of growth hacking. Which, hey, is also awesome. Some folks are calling this the work of GTM engineers. (A marketing technologist by any other name would smell as sweet. Um, metaphorically speaking.)

But as enthusiastic as I am for marketers and marketing operations pros to be imbued with these AI superpowers, I feel like this framing misses the spirit of what vibe marketing really should be about. Production agents, apps, and automations don’t seem vibe-y to me. They seem more like your marketing operating system. You want it to be fluid and adaptable. But you also want it to be structured, reliable, and well-governed.

My interpretation is that Andrej’s good vibrations with vibe coding came from the freedom to think less about code and more about indulging one’s imagination to bring new ideas to life, even as imperfect prototypes. The “vibe” is human creativity unleashed.

That is the vibe that marketers should embrace.

Better yet, marketers should be the “market whisperers” and “customer whisperers” who seek to tune into the vibes of their audience. If vibe coding delights developers — they’re the ones feeling the vibes — vibe marketing should focus on delighting customers.

Auomated content generation. Automated outbound emails. Automated data scraping. Don’t get me wrong, these are all useful AI-powered capabilities that marketers can and probably should implement in their operating system. But the mechanics of this aren’t the vibe. The real question is how prospects on the receiving end of that machinery feel about it. Are they humming The Beach Boys’ Good Vibrations or Monty Python’s Spam, Spam, Spam?

To me, vibe marketing should be about experimentation.

Curious about a customer pattern? Use AI to mine your data lakes and warehouses for answers that previously would have required an analyst or data scientist — a high enough barrier that the vast majority of questions that popped into one’s head weren’t worth pursuing. But now? Let those “I wonder…” curiosity vibes flow.

Those answers lead you to a hypothesis? Compose a micro-campaign, using AI to develop the creative — which can have more ambitious, app-like, multimodal, and dynamic concepts than were ever previously in reach for an individual marketer to whip up solo — segment a test population, deploy, monitor, and measure against your hypothesis. Most experiments should be quick, cheap, and low-risk. Iterate the promising ideas. Scale the proven ones. Jettison the duds.

Which experiments vibe with your audience? Those get promoted from vibe-land to a more polished production plateau.

If Big Data was about harnessing the volume, velocity, and variety of data, vibe marketing is about Big Experimentation — harnessing the volume, velcoity, and variety of marketing experiments to out-innovate your competition.

Yes, there need to be guardrails. Proper permission management for data. Compliant respect for customer privacy and preferences. Adherence to brand standards. But this is increasingly manageable with professional AI martech tools. One of the greatest contributions marketing operations teams can make in this environment is setting up the scaffolding and safety nets for the rest of the marketing team to vibe more ideas to life.

Will the “vibe” language endure past this 2025 Summer of Vibes? Probably not. It’s a little hokey. But the creative power of Big Experimentation unleashed by a new generation of AI-empowered marketers is only going to grow — whatever label we stick on it.

For now though, enjoy the Good Vibrations of summer.

tl;dr The martech landscape grew again since last year, up 9% to 15,384 solutions. While new AI natives continue to blossom, the previous generation is consolidating more. There’s also an exploding “hypertail” of custom software. We cover all of this and more in our new State of Martech 2025 report (free and ungated).

Over 15 years ago, I tossed a bunch of martech logos on a slide to present the Rise of the Marketing Technologist to a room of CMOs, making the case that marketing was becoming a very technology-integrated discipline. (For our younger readers: “marketing technologists” were what we called go-to-market engineers back in the day — you know, before electricity and indoor plumbing.)

The next year, I formalized the slide into the first official marketing technology landscape, with a whopping 150 — gasp! — solutions categorized. From there, things got surreal. The market of martech grew exponentially to thousands and thousands of products. Every year, pundits predicted it would collapse within the subsequent year. And every subsequent year, empirical reality proved otherwise.

Which brings us to today’s release of the 2025 marketing technology landscape, now with 15,384 solutions organized in 49 categories. Yes, it grew. Again. Up 9% from last year’s 14,106. Massive credit goes to Frans Riemersma, who has been co-producing the landscape with me since 2022 with a small army of researchers and volunteers.

Of course, as a mere slide, it’s been completely unreadable for years. But to prove that it’s not just random dots on a page, you can:

• Download a hi-res, clickable PDF of the full martech landscape
• Visit MartechMap.com to zoom, search, and filter an interactive version

“If you keep squinting at all those little logos, you’ll go blind!” I’m sure my mother would have advised. So let’s step back and look at the growth quantitatively, year-over-year. The headline — literally — is that the martech landscape has grown 100X since 2011. Or 10,156%, which seems larger. (Yeah, I’m bad at math.)

Death of a Martech Salesperson? For some, yes

Now, before you accuse us of sandbagging this by including anything thrown our way with a URL and a favicon, I’ll note that we actually evaluated 11,133 new candidates this year, out of which only 2,489 qualified. At the same time, we removed 1,211 from last year’s landscape that vanished either through acquisition or simply shutting down. That’s a churn rate of 8.6%. (Again, being bad at math, the 1,211 number feels larger to me.)

You might expect that most of those that went away were flash-in-the-pan products from last year’s massive explosion of new AI-native products that surged after the release of ChatGPT. In fact though, 2/3 of the ones that went away were from the previous decade, 2010-2020.

Yes, the consolidation of that first big wave of martech products is finally happening. The laws of gravity can be defied only so long. While many of the ones that disappeared were smaller, organically funded solutions in the “long tail,” the more newsworthy ones had raised significant venture capital as the darlings of their time.

A toast to them from Four Weddings and a Funeral, “May we all in our dotage be proud to say, ‘I was adored once too.'”

Yet the great circle of life martech continues…

But that hasn’t dissuaded others from diving in. With AI further lowering the barrier to entry for building software — while simultaneously widening the vista of new opportunities for innovation — martech startups continue to be born at an unrelenting rate.

Last year, there was an incredible explosion of new startups in content and sales, at 35% and 47% growth rate respectively. These were two domains that immediately benefited from the conversational magic of generative AI.

This past year, growth has been somewhat more subdued — and more evenly distributed — with ~7-10% growth in each of our 6 major martech landscape domains.

Digging down into subcategories, it was interesting to see SEO had the highest percentage growth of all — an eyebrow-raising 24% — albeit more modest in absolute numbers, expanding from 212 to 262 products. It’s ironic, given all the “SEO is dead” talk of the past year. But in martech as in life, when one door closes, another one opens: there are new product to help you optimize your content for AI services, AI Optimization (AIO), which we’ve lumped into this subcategory.

With the continued proliferation of new products, especially AI-native ones, the average tech stack grew a little last year, by 2.2%. For more mature marketing teams though, the fluidity of their toolset has been balanced by aligning and integrating these new AI products with an established platform at the center of their stack.

We re-asked the question of which platform martech and marketing operations leaders considered to be the center to see any shifts from 2024 to 2025. In B2B companies, there was very little change: most consider either their CRM or their marketing automation platform (MAP) to be the center.

One notable change in the B2B segment was the 5X rise of “Other” as an answer, from 2% to 10%, which includes custom-built martech platforms. While that’s still a small slice of the pie, it’s an interesting reversal of a decade-long shift from custom to commercial martech.

With B2C and hybrid B2B/B2C companies, the most notable change was with CDPs, which dropped from 26.9% to 17.4% of those surveyed citing them as the center of their martech stack.

CDPs appear to have been squeezed between two layers of the stack. At the data layer, we saw cloud data warehouses rise as the identified center platform, from 20.9% to 23.9%. At the engagement layer, we saw marketing automation platforms (MAP) and customer engagement platforms (CEP) grow from 19.4% to 26.1%. CRMs also grew from 17.9% to 19.6%. CDP capabilities seem to be moving to either the warehouse layer (“composable CDPs”) or the engagement layer, where they are increasingly embedded into customer-facing platforms.

1,000,000X more martech that’s out of sight

But arguably the most fascinating development in martech isn’t the long tail landscape of commercial products. It’s the rapid proliferation of custom-built apps and agents, what we call the hypertail.

There’s always been the option for businesses to build their own custom software solutions. For most companies, that represented a small amount of their tech stack — if any at all — because custom software was expensive to develop and maintain, requiring engineering skills that weren’t readily at the disposal of most marketing teams.

But the rise of low-code/no-code platforms over the past 5-10 years has steadily changed that equation, making it easier, faster, and cheaper to build, giving rise to “citizen developers” who could increasingly scratch their own itch by creating lightweight apps and automations.

As with nearly everything else it touches, AI is having a massive multiplier effect on this. It’s accelerating software creation at a blindingly fast rate. Professional developers now rave about “vibe coding” — whipping up new programs in a matter of hours that used to take days or weeks to construct.

For citizen developers it’s an even bigger boon. If they can describe something in a natural language prompt, the AI can build it for them. True, today that works best for simple apps and automations. But this is Clay Christensen’s classic pattern of disruptive innovation: serving the underserved with simple use cases, and then steadily climbing up the hill to more advanced ones. In our last report, we called this “instant software” — just add ideas.

And it gets wilder. Popular AI assistants such as ChatGPT, Claude, and Gemini — that have millions of non-technical subscribers — are often creating software programs behind the scenes to do the bidding of their users without those users even knowing that software was built and run on their behalf.

The result? Not millions but billions of custom software programs proliferating companies’ digital operations. Probably trillions, many blinking in and out of existence on demand.

That is the hypertail.

Download The State of Martech 2025 for free

I’ve just touched on a few of the many findings and insights from the full State of Martech 2025 report, which is available to download, free and ungated.

We want to thank GrowthLoop, Hightouch, MetaRouter, MoEngage, Progress, and SAS for sponsoring our research. We also want to thank Goldcast for providing the events platform for our #MartechDay sessions, which you can still watch on-demand for the month of May by registering at martechday.com. And we also thank G2 for sharing ratings data with us for our analysis of the martech landscape.

This is a short excerpt of data from The State of Martech 2025 report that will be released on May 6, 2025. You can get the complete 100+ page report and access to an extended keynote presentation of our research by registering for free for #MartechDay.

It’s no surprise that stand-alone AI assistants, such as ChatGPT, Claude, and Gemini have become ubiquitous. In our study of 96 martech and marketing operations leaders, 87.5% reported these tools are now widely used in their marketing department.

But what about deeper embedding of AI in the martech stack?

Overall, 42.7% of our respondents said that AI tools now integrate well — or with only minor challenges — with their existing martech stack. Although we saw a significant difference between B2B vs. B2C companies, with 54% of B2B reporting easy integration compared to only 15.4% in B2C. (We note that this is an inversion of a prior stereotype of B2C companies being more advanced in martech than their B2B peers.)

So what are marketers doing with these integrations?

We started — as all good AI discussions should — by asking about data. In particular, unstructured data, as this is where generative AI LLMs have unlocked significant new value. So many insights were buried in calls, emails, and other conversational interactions with customers. But until recently, tapping into those insights in a programmatic fashion was difficult, slow, and expensive.

Suddenly, it’s now quite easy.

65.6% of our respondents said they currently use AI to capture, analyze, and use unstructured customer data. The most common unstructured data sources are call recordings and transcripts (41.7%), emails (26%), and chatbot transcripts (26%).

A fascinating source of unstructured data that’s emerging, primarily in B2B, are customer websites. 15.6% of our respondents said they scrape websites of prospects and customers to learn about them and tailor messaging and engagement accordingly. Instead of often outdated third-party data about a company, you can now get it straight from the source. This will be game-changing for account-based marketing tactics.

The integration of LLMs into marketing workflows and automations enables a number of creative and productive use cases:

68.7% of the companies in our study have at least one LLM workflow or automation already implemented. “Agentic workflows” are what they’re popularly called, even if the agentic-ness of them — how much autonomy is actually in play — is sometimes debatable.

The most popular use case is summarizing messages and content (38.5%), followed by personalization of messages and content (31.3%) — essentially working with unstructured content flowing in and out of the organization. 22.9% use it for translating messages and content in either direction.

25% use it to research customers/prospects to enrich their profiles — again, more common in B2B environments.

It’s the 20.8% who use LLMs to make or recommend a decision based on structured and/or unstructured data provided that can claim to be actually implementing agentic workflows. Such intelligent decision-making with at least some degree of autonomy has tremendous potential to advance the sophistication and effectiveness of marketing operations. But it’s not surprising that companies are proceeding carefully here. This is powerful, but very new, and we need to feel our way around the jagged frontier of what’s reliable.

Finally, where are these AI automations and workflows being built and run?

Earlier this year, we discussed the seeming free-for-all around digital ops orchestration with AI. Existing SaaS marketing platforms and popular iPaaS/workflow automation products are all vying to be the conductors of AI automation — essentially just expanding on their current, pre-AI role. But they’ve got competition from a wave of new AI agent platforms, as well as more and more custom-built AI.

The results of our study show that, at least for now, the incumbent SaaS marketing platforms — CRM, MAP, CEP, DXP, CDP, etc. — have the lead, with 44.8% reporting that’s where they’re implementing AI automations and workflows. The next two, iPaaS/workflow automation tools and custom-developed AI, have less than half that adoption each, at 19.8% and 18.8% respectively.

New AI agent platforms — such as CrewAI, Lindy, Relevance AI, SuperAGI, etc. — have only 8.3% adoption.

Of course, this is still the first inning, if not the pre-game warmup. But initiative matters, and currently the SaaS platforms that have been the backbone of the martech stack are the ones that have it.

This post barely scratched the surface of all the data and research analysis included in the full State of Martech 2025 report. Don’t miss the opportunity to get the full report — along with the latest martech landscape, a deck of illustrated martech stacks from companies such as Itaú, Nationwide, Thomson Reuters, and Verizon, and access to an in-depth presentation of this material live or on-demand. Sign up free for #MartechDay now.