DHS

Startup investing heuristics

Tue, 14 Apr 2026 22:49:00 +0100

Here are some of the factors I weigh up when deciding to invest in a startup.

Green flags

- High pace of execution
- Ambitious
- Technical cofounders
- Great hustle / strong discovery process
- Strong early hires
- Growing fast
- Targeting US market
- Potential to build a $10bn company
- Ramen profitable
- Someone I know
- Or a friend of a friend
- And they’re excellent
- Previous (significant) exits
- They don’t need my money
- Angels I respect
- Going through YC
- Picky post-YC VC (eg. Rebel fund, Standard Capital)
- S-tier VC
- Need capital to meet demand
- Building a future I want to see
- Fit with an investment thesis I have
- Can take small cheques (via syndicate or RUV)

Red flags

- Company has been going a long time
- Biotech
- Extension / bridge rounds
- Flat graphs
- Same firm leading multiple rounds
- Cashflow crunch
- Team are senior execs not founders
- No investors I recognise
- Need money for MVP
- Backed by insiders / spin out

Investment thesis: Continuous Everything Monitor

Mon, 01 Sep 2025 10:00:35 +0100

Imagine having an instant picture of every biomarker in your body.

Whenever you’re feeling a bit off you can see exactly what you need to feel better again. Not just ‘is my blood sugar low?’ but 'have I eaten enough protein in the last 24hrs?’ and 'what are my hormone levels right now?’

I see this as completely inevitable given a long enough timeline.

We’re seeing the first glimpses of this already of course: Apple Watch reporting heart rate / HRV / sleep, Whoop doing the same and recently added blood pressure & ECG, Continuous Glucose Monitors are getting smaller. Function Health is making comprehensive blood panels cheaper (and annual MRIs via recent Ezra acquisition). There are also services to track cortisol (Hanu Health) and your microbiome (Zoe). Piece by piece the puzzle is coming together.

There are many breakthroughs required to bring the full picture version of this to life.

And yes there’s an argument that in some cases more information isn’t always better but that’s a choice people can make. The data will be available.

I’ve seen a few decks recently that align with this thesis but have yet to make an investment. Do let me know if this resonates with you or if you know any teams working on this.

Discovery machines

Mon, 25 Aug 2025 14:26:22 +0100

Discovery is the process of identifying pain points, testing demand / willingness to pay, developing an MVP, and thus building momentum towards product-market fit.

Most founders have an idea, build it, then do a half-hearted retrospective discovery process and close the company if it doesn’t work out. Perhaps with a pivot or two in the dying breaths just in case something sticks.

There’s a better way: an up-front discovery process. Fixed on the goal of building something valuable but flexible on what that might turn out to be. Without wasted up-front building. This significantly reduces risk of failure.

Some later stage startups continue with this even after finding product-market fit with their core product. See all the categories Revolut is building successful businesses in. And Cleo explicitly hiring Discovery Engineers (role shared here last week).

I believe founders building discovery machines offer a superior risk profile to themselves and to investors 👀

Early stage VC is already a way of placing multiple startup bets - but if every portfolio company is using this approach they get so many more bets.

Side note: through this lens it’s fascinating to think that Google & ChatGPT both have users typing their pain points straight into a textbox.

PS. As always, caveats abound. Yes, some experienced founders are acutely familiar with the pain points and can just start building (eg. Immad Akhund @ Mercury, Parker Conrad @ Rippling). Yes, this still requires capital, energy, discipline, speed etc.

Why UK angel investors should invest in Silicon Valley startups

Fri, 22 Aug 2025 07:28:27 +0100

Highest density of capital, talent and breakout companies in the world
Investing is a global marketplace - there are no bonus points for limiting yourself to your local market
The best startups should be there and if they’re not we should be encouraging them to have a presence there to increase their chance of success
Revenue graphs from Silicon Valley startups look like the fantasy projections of other startups
If you make money doing this you can reinvest in local startups
Of course there’s froth and higher valuations
Outsiders can get access via AngelList syndicates

Unwritten rule of startup fundraising: there’s a hierarchy of capital 🪜

Thu, 21 Aug 2025 17:27:34 +0100

The best entrepreneurs raising VC start by approaching the top tier funds.

If they don’t bite then they approach 2nd tier funds. Then third tier / regional funds.

If none of those bite then crowdfunding is the next option.

This is why the performance of the top VC funds is so far ahead of the rest - because they get to pick from the best startups.

The best hope for all the other funds is either that the top funds overlook a great deal, or that they can build a meaningful/distinctive value-add or special relationships of some kind.

And of course there are always exceptions 😉

Cande y Paulo - Barro Tal Vez (Luis Alberto Spinetta)

Fri, 21 Apr 2023 09:43:55 +0100

Cande y Paulo - Barro Tal Vez (Luis Alberto Spinetta)

The Small Net

Mon, 02 Aug 2021 13:58:15 +0100

I think I first got online around 1996. And I miss that early Internet. It was a simpler world of weird websites and individual blogs. Nowadays the Internet is dominated by addictive social media, political mobs, an ocean of e-commerce, and pernicious advertising.

And web design has evolved beyond recognition too. In the early days it was common to hand-roll html files. Now there are a plethora of frontend and backend frameworks to choose from, each with an iceberg of dependencies.

I’d heard of a movement called the small net (aka smolnet or small web) as an antidote to all of this but it hadn’t quite clicked.

I recently found an entertaining and intriguing article proposing PDFs as the antidote to complex modern web design. I emailed the author in appreciation and he tipped me off about a project called Gemini. Suddenly the small net clicked…

The Gemini protocol is designed to be a simpler equivalent to http. It exists in a parallel universe to the Internet referred to as gemspace. Websites on Gemini are known as capsules and their addresses start with gemini:// instead of https:// You can’t browse gemspace natively through a normal web browser - you have to use a special browser or a proxy. Check out Wobbly from a web browser, Elaho on iOS, or portal.mozz.us.

Gemini markup is constrained to titles, text, links, bullets and quotes. There is no other formatting and no CSS. Images and links can’t be rendered inline with text. Links must be on their own line. And images can only be linked to. Luckily emojis are supported 😉 These constraints result in a focus on content and creativity.

It’s common for capsules to include a gemlog - equivalent to a blog - and to link to other capsules that their creators admire. This makes browsing gemspace a very personal experience - which is exactly the intention.

Browsing Gemini feels like the early Internet and I love it. Come and explore Gemini with me. And if you’d like a head-start on creating your own capsule, check out flounder 🐟

Defining (artificial) intelligence

Tue, 13 Jul 2021 14:54:58 +0100

I was astonished by GPT-3 when it was released last year. It was a machine learning system that could apparently write entire paragraphs quite coherently.

Shortly afterwards a discussion with an AI researcher friend made me re-think. If you use a large corpus of training data to guess which words might come next, is that really intelligence? Does it ‘understand’ what it’s writing?

(This notion became one of the threads that wove into my recent project Pan.)

A few weeks ago I was discussing this with a different AI researcher from DeepMind. Their response was to ask whether humans always ‘understand’ the answer they give to questions? Our answers often come from repeating something we heard elsewhere, guessing or just not properly thinking something through. Are we really so superior to a machine learning system guessing the next word based on a large dataset?

Last week I heard Emily Bender, co-author of ‘On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜’ take part in a panel discussion. Her take on this point was very different again. She said that to imply that machine learning are in any way equivalent to human intelligence is to fundamentally deny our humanity. She believes this so strongly that she refuses to debate with people who take this view!

Crypto #3

Mon, 05 Jul 2021 09:17:56 +0100

Previous posts here and here.

Since first buying ETH (and doubling down in a dip) I’m now up ~10X. And that touched 20X a few weeks ago.

Confidence is a determinant of the strength of a currency. When crypto was a novelty, confidence was low. As large institutions and sovereign wealth funds allocate to crypto, confidence is accelerating.
El Salvador recently became the first country to adopt Bitcoin as legal tender.

Previously I had thought of Coinbase as the #1 crypto exchange. And said that investing in Coinbase would be a good proxy for investing in crypto. Since then Coinbase went public at a valuation so high it has put me off.
I’ve also learnt that by volume Binance is far bigger than Coinbase. They are however a more controversial player. They’re based in China where crypto is banned(!) and some of their products are banned in other countries too. The UK’s FSA recently clamped down on their options trading service. Last week it wasn’t possible to buy crypto with GBP on Binance though this this has since been resolved.
I’m a fan of backing the #1 player in a consumer market as so much extra benefit accrues to the leader. Binance shares are not publicly traded but there is an interesting proxy available in their own in-house coin. BNB is aiming to be a smart contract platform like Ethereum. It’s currently trading at 50% off its all-time high (ATH) so I’ve bought in.
Andreessen Horowitz who I’m a fan of and are very bullish on crypto, recently made a large investment into a platform called Solana. Solana is another smart contract platform. Currently at 50% off its ATH, I’ve bought in.

A new dimension to me in crypto is staking. This means tying up your coins for a period of time in exchange for a reward.
The main risk is that if the price of a coin crashes then you can’t sell. With my long-term, conviction-led approach, the price crashing is an opportunity to buy more so - apart from extreme cases - this risk is mostly irrelevant to me.
The most significant staking project is the upgrade to Ethereum 2.0. This will dramatically increase the speed and energy efficiency of the Ethereum network. The end date of this upgrade is unknown but estimated to be 1-2 years away. The staking reward is currently equivalent to ~6% (decreasing as more coins are staked). Coinbase don’t offer ETH staking but Binance do so I’ve used a proportion of my ETH for this.
Binance also offer 30 day staking in BNB and SOL so I’m using that too.

Thanks to my OG crypto trader friend for the discussion & advice!

Dom Whiting’s ‘Drum & Bass On The Bike’ sets are a great...

Fri, 02 Jul 2021 15:53:01 +0100

Dom Whiting’s ‘Drum & Bass On The Bike’ sets are a great combination of:

Great tunes, well mixed 🔑
Livestream tech: bike, decks, camera, mic, 4G
Livestream interactions: shout-outs to online viewers
Real world interactions: followers on bikes, amused pedestrians, cars honking
Just a good dude with an infectious passion!
Capturing the public’s imagination: this London set now has 1m views and hundreds turned up for his recent ride round Manchester

I stumbled on Jacob Collier’s reharmonisation of ‘Fix You’ a few...

Wed, 23 Jun 2021 16:34:13 +0100

I stumbled on Jacob Collier’s reharmonisation of ‘Fix You’ a few months ago and found it fascinating.

A commenter on YouTube captures my thoughts:

It’s weird, I watch this guy and I’m not even that crazy about his stuff, but his ability - in every sense; vocally, instrumentally, compositionally, theoretically - is so superior, that I’m completely fixated.

There’s something about his manner (talent? confidence? enthusiasm of his fans?) that I find a little embarrassing for some reason. But his talent is unmistakeable. Check out:

Instagram Live explaining harmonisation
Charles Cornell on Why Musicians Love Jacob Collier
Playing Sparks with Chris Martin
Musical conversation with Charlie Puth
His epic Grammy-winning arrangement of Moon River

I’m hooked on Marc Rebillet’s positive energy and musical...

Wed, 23 Jun 2021 12:33:10 +0100

I’m hooked on Marc Rebillet’s positive energy and musical talent. His Sunday livestreams are just the ticket!

(Thanks for the tip Mark!)

Etsy acquires Depop

Thu, 03 Jun 2021 11:00:30 +0100

Etsy acquires Depop:

I mentioned Depop here back in October 2017. Today they announced their acquisition by Etsy for $1.6bn. A huge result for a UK startup and will hopefully have a ripple effect for other startups too.

Juggling From Above (via kottke)

Wed, 02 Jun 2021 09:40:33 +0100

Juggling From Above (via kottke)

New Sigrid!

Tue, 25 May 2021 18:48:20 +0100

New Sigrid!

Do you have enough time to take on a new project?

Tue, 25 May 2021 18:38:08 +0100

Do you have enough time to take on a new project?

Proud Virgin Galactic investor today! 🚀

Mon, 24 May 2021 12:49:45 +0100

Proud Virgin Galactic investor today! 🚀

John Berger’s Ways of Seeing

Sat, 22 May 2021 14:51:45 +0100

John Berger’s Ways of Seeing

The neuroscience of napping

Fri, 21 May 2021 19:10:38 +0100

The neuroscience of napping

Guardian » Documentary films » Gunda

Fri, 21 May 2021 13:08:32 +0100

Guardian » Documentary films » Gunda:

Sensationalist headline aside, this sounds like it’s going to be a beautiful film.