Actionable BI

At a recent conference, I had an opportunity to interact with a lot of clients and I realized there is a lot ambiguity and confusion about Actionable BI. While there is a lot of demand and appetite for it, there is a lack of clarity on the subject… So I thought of exploring Actionable BI through a series of blogs to share my thoughts on it. Here is the first post in the series that will first introduce what is Actionable BI for you. In the next three posts, I’ll attempt to discuss:

• How to achieve Actionable BI
• How do we make data for Actionable BI trustworthy?
• Role of AI in making data for Actionable BI Trustworthy

Organizations spend millions of dollars on producing thousands of BI reports that may provide operational or historical data to show past trends and spikes. But business users need to do a lot of investigations through drill downs and additional reports or queries to make them actionable.

Difference between BI and Actionable BI
Techopedia defines actionable data or insight as “information that can be acted upon or information that gives enough insight into the future that the actions that should be taken become clear for decision-makers.” So while Business intelligence (BI), is often described as “the transformation of data into relevant and useful information for the purpose of business analysis” the term Actionable Intelligence is defined as “information that can be acted upon, with the further implication that actions should be taken.”

Characteristics of Actionable BI

1. Drillable and On Demand – It should be available when you need it at the point of interaction with customers, partners and other stakeholders in the complex supply chain and ecosystem so that business users don’t have to go to a separate report or dashboard. It should also be easy to drill down to the level where the BU report not only highlights the area that needs attention but take you down to the level so that you know exactly which customer or supplier you need to call and why.
2. Contextual – It should know the context in which you need the information e.eg if you are making a decision about an ordering a part from a supplier, it should know suppliers’ past performance about delivering the part in terms of quality, timeliness. It should align that with demand for the part and tell you if the part is likely to arrive on time.
3. Integrated with business process – Actionable BI is really effective when business users don’t have to jump through hoops to get to it. If it is integrated with the application or business process so that it can provide contextual business intelligence on demand. In the above example, if BI about the supplier performance is integrated with the procurement system, users will have the intelligence available without having to log into another BI system.
4. Smart Alerts – Smart Alerts allow users to receive critical business information in the quickest and most efficient possible way. For example, a store manager can be automatically informed when in-stock levels of critical items fall below or rise above a certain level.
5. Augmented Intelligence – Augmented Intelligence can analyze huge amounts of data generated from machines, GPS devices, and other IoT sensors to glean insights and provide human experts to take appropriate actions
6. Constant Feedback Loop – For the system to provide accurate and reliable intelligence for effective decision making, BI systems need to stay up to date through constant feedback loop based on the results of the actions taken based on the intelligence provided. I designed a system for an aircraft engine repair company that allowed them to recommend specific actions to optimize the cost to repair the engines. Shop floor technicians would track the recommended actions and provide feedback to the system on the new estimated cost to repair the engine.

The post Trusting and using Actionable BI appeared first on SogetiLabs.

Technology is impacting the way we behave. It’s about the way we communicate, the way we make decisions, the news we read, how we do business. Our smartphone is an extension of our body and our minds. We use it as a tool to speak, listen, look, think and decide. It is not so much about technology, it is all about human behavior. We see the digital world and the natural world becoming more and more entwined. The same should happen with experts in different disciplines.

The questions which arise in current society, related to usage of technology, are very diverse:

• what does the usage of smartphones mean for our physical and psychological well-being?
• what will the impact on our democracy be when we all live in our own bubble where we only read news which confirms our opinions?
• what kind of ethics should be implemented in artificial Intelligence software and how do we govern that?
• how do we keep our lives meaningful when robots take our jobs?
• how will we feel and behave when camera’s and microphones are open always and everywhere?
• what is the impact on our communication when we are never sure if we are talking to a human, a robot or something in-between?

Image source

The list of important questions is endless.

These questions are not the expertise of  IT-people so we should not leave it to them. Regularly I hear IT colleagues say that indeed it would be good to involve other disciplines in what we are doing. That is a positive thing but by far not enough. It should not be the situation that  IT develops things and consult people from other disciplines to ask what they think of it. The tech world should really open up and cooperate with people with a background in philosophy, psychology, ethics, sociology, politics, design, and arts on at least equal level. Or even better, those are the people who should take the lead.

Technological experts have the natural need to invent and develop new things and make it technically better. But the fact that it is possible to produce something, does not automatically mean that it must be produced. Technical experts are the worst people to determine and keep those boundaries.

It is not without reason that Cathy O’Neil named her book “Weapons of Math Destruction”. It is about the impact of big data and artificial intelligence. The title relates to “weapons of mass destruction”, which is one of the very few inventions for which mankind decided that there should be a global restriction on production. I foresee a similar need for certain artificial intelligence and we desperately need the “soft sciences” to take the lead in those discussions.

Image source

The post Technology is way too important to leave it up to tech people appeared first on SogetiLabs.

I’m currently in Tel Aviv and want to share 3 takeaways from the Future of Learning conference I just attended. It was mainly about AI and Education. There were some really interesting speakers such as the Minister of Education of Israel, and Gabi Zodik, CTO IBM Watson, accompanied by a lot of smart PHDs and doctors.

Watch my short impression of the conference in Israel.

The post The Future of Learning: Three takeaways from Shaping the Future in Tel Aviv appeared first on SogetiLabs.

Based on their recently launched new book “Testing in the Digital Age – AI Makes the Difference”, Tom van de Ven and Rik Marselis explain the 5 Hops to digital testing success with Artificial Intelligence.

Watch the webinar to learn more.

The post [Webinar] Testing in digital age; AI makes the difference appeared first on SogetiLabs.

I am increasingly being met with stories about what Digital Natives can do and how they will influence the work place and society in general.

It is true, that you may need new and different incentives to retain generation Z in the workplace, and that is of course an influence…

In most of the developed world we are seeing a shrinking workforce and many companies will strive to attract employees with a broader background than usual – which also includes work immigrants, and people over the usual retirement age.

Organizations should strive to implement incentives that meets the individual’s demands no matter their background and not try to stick to standard policies for all to retain a diverse workforce in an exciting an innovative environment which is needed in a world of accelerating change.

Changing the organization

You should strive to employ a diverse workforce on all aspects to create the most productive and stimulating environment. With the demographics changing from a pyramid to a spinning top in shape, you need to cater to each individual employee’s need in a personalized fashion to make everybody feel at home and welcome in the organization. That includes both the iGeneration employee who will instinctively know how to swipe the factory floor with a broom and the baby Boomer who will have the experience to create the next generation SmartHome control. It also includes both the workaholic happily spending 60 hours or more a week at work and the parent of two kids who prefer to spend as much time as possible with family.

Experience is just as important in innovation as is open mindedness – actually, many innovations are all about applying old knowledge in a new context and in the process create new knowledge. Having worked with assembler programming, sensor readings and 300 BAUD modems in the 80’s is a very useful experience in today’s IoT world – it’s about reliably communicating data despite various issues and limitations.

Being from a specific generation – called Digital Natives – is not a good definition of people who will have a digital impact on the workplace or society in general. I have colleagues that have passed 60 (years of age!) and who are more digital savvy than many just-out-of-college teenagers.

Who Are the Digital Natives?

“The media has focused a lot on millennials in recent years, but it’s time to turn some of the attention to the millennials’ future co-workers. Gen Zers have a lot in common with millennials, but there are also many ways in which the two generations differ”. These words are from Deep Patel in an article at forbes.com

Though the eight overall trends of this generation described in the Forbes article are correct. This generation has just as big a variance in personality and intelligence as any previous generation.

Bear in mind, that the term Digital Natives is most commonly used about generation Z (people born mid-90’s to mid-2000’s and also often called the iGeneration) and sometime also for generation Y (born mid-80’s to mid-90’s and also often called Millennials) and has no link to skills or other qualitative factors. Some say that it should refer to people grown up with ubiquitous access to WiFi – but that would exclude people growing up in many places today as I experienced at a recent visit to Edinburgh, Scotland!

To give an example of a Digital Native that is absolutely lost with everyday technology: I know a recently graduated Master of Business Administration, mid-twenties, who doesn’t get the concept of booking a meeting with two other persons in a meeting room using Microsoft Outlook, and who is not able to combine selected slides from two different PowerPoint slide decks into one presentation in a common layout – two very common tasks in many white-collar workplaces. But who knows how to swipe and pinch! When offered a course, that person turns it down as not necessary. That’s going to be a very long learning curve to productivity in the workplace.

The post Digital Natives appeared first on SogetiLabs.

Based upon my experience of 5+ years on Azure and 1 year on AWS here I am sharing my view on major factors & cloud offerings which influence the cloud provider selection decision for any customer.

Organizations are looking at AWS and Azure as primary cloud providers and taking a cautious decision while they build new systems or migrate their existing landscapes to clouds. AWS started the cloud offering in 2006 well before any other cloud providers and has a very strong capability on the Infrastructure side. Microsoft started its cloud offering in the year 2010 and has grown very fast to become the second largest cloud provider after AWS. Microsoft is already running its software’s and products in most of the client portfolios which are easier to migrate to Azure due to its native support to all these systems. AWS, being an old cloud provider has a global reach and maturity in its offerings which is, the main reason for hosting and running business critical production systems on it. There has been a trend in the market where few customers choose a cloud initially and then make their mind to change based upon the comparison on below factors depending upon which services are more relevant for their scenario and who has an edge in that offering:

Compute

Compute capability of a cloud provider is measured in terms of a number of compute instances, number of CPUs, memory size, OS supported along with a number of images available in the market place on it. AWS has an edge on Azure in terms of price when it is compared cumulative for CPU, RAM, and SSD storage, there might not be exactly one to one mapping of machines from both the clouds but still, AWS is reasonable in this area.

Data Persistence

Storage is one of the critical offering needed by the clients to reduce their cost to maintain data backups at on-premise infrastructure. Azure offers Blobs and AWS has S3 to cater to this need of the customers. Main features which are needed to ease the use of these offering are batch imports of data, data encryption, redundancy reduction and size of the blob where again AWS has an edge on blob size, cost of persistence when data is in TB and auto archival & deletion feature which saves a lot of cost for customers.

Database is another main feature which is needed by all the applications irrespective of their size and business criticality. Azure and AWS both have a major focus in this area let it be RDBMS, NOSQL and Data Warehouse. Both cloud vendors are equally good in the features but again AWS wins barely in terms of maturity.

Support & Troubleshooting

AWS has a larger market share as compared to Azure or any other clouds. AWS does more conferences and webinars throughout the year and has larger support team and a quick turnaround which is a plus point for the application & infrastructure management teams.

Identity Management

Azure and AWS both have similar offerings for security controls at services level and infrastructure level. When it comes to applying role-based access policies for their access and restriction policies AWS is mature here reason being its long-time existence in the market.

Release Management

Azure and AWS both support the tool and script-based deployment but the features like natively integrated release orchestration platform i.e. VSTS, ease of integrating code analysis, functional testing tools and options like Hot swapping of releases gives Azure an edge over AWS.

Ease of Access

Both the cloud providers have similar compatibilities for console-based and web-based access. Azure is much easier to access and rich portal view as compare to AWS, so Azure has an edge here.

Conclusion

Azure and AWS both the cloud providers hold a big chunk of total market share in terms of hosting the customers’ workloads on cloud. Both have an edge over one another for different services and cloud being in its adoption stages, so most of the customers are in process of implementation of cloud first approach which requires maturity, support and global reach for the areas like compute, storage and support. AWS is matured in all these areas and cost-effective as well, that’s why traction for these kind of migrations is toward AWS. Also, for some customers, these factors might be a driver from Azure to AWS migration.

Gartner has published the below updated cloud IaaS scores for Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP) in March, 2018:

For a detailed study of the Gartner report please visit https://blogs.gartner.com/marco-meinardi/2018/03/13/just-published-new-assessments-aws-azure-gcp-cloud-iaas/

The post Which cloud, best suit my needs – Azure OR AWS appeared first on SogetiLabs.

I have recently been confronted with some interesting innovations on different clients’ projects, showing, besides the benefits, some similar issues. One case with a mainly technical innovation, the other with a complete digitalization of a process previously delegated as administrative tasks (digitalized as well).

Innovation is often a good thing, sometimes even key to grow or to maintain a reasonable cost compared to competitors. It’s always an interesting experience as it “shakes” our habits, gives us new options, and offers new points of view on activities we have engaged on, sometimes for decades.

Innovating is etched in our practices, a great part of what we do as architects is making possible and supporting new and improved processes, products and interactions for the different stakeholders we work for, may they be businesses or IT.

But innovating can become an habit in itself. Especially the improvement innovation. Using a more robust solution, a more powerful technology, adding some new “best practices” to maintain a better service or faster operations. But in those cases, we were integrating some disruptive innovations.

Interestingly, we didn’t have many issues with the innovations themselves, people did a relatively great job implementing them, they were well received, the breakthrough they were to deliver was quite clear. Most issues were caused by considering that what we did previously, processes and technologies improved during decades, were still relevant. Expecting the problems and solutions to be similar, using the same processes or practices to respectively reduce and build them. What became clear was that we had to rethink some of the solutions (the previous innovations) some considered as a given.

By innovating, in those cases, we changed where we were in the solution space. Giving us many new and improved options, cheaper, faster, and simpler. But we also lost the iterative improvement done on past-solutions, we also broke some processes most actors considered imperative. We lost months combating inadequate processes or certainties that were not relevant anymore in this new solution space. Worst, a lot of actors spent a great deal of time to try to rebuild those processes and solutions as they were.

Perhaps a lesson would be to spend more time during the first phases of a project on the disruption embarked by using innovative solutions, differentiate them from improvement and identified with the different stakeholders what those innovation will change for them and what processes and solutions need to be adapted or used more carefully.

In both cases, the result remained positive, but obviously, time could have been better spent and more enjoyably improving those solutions instead of fighting them in the wrong space and slowly aligning most of the actors. In that dimension as well, digital happiness could be an inspiring dimension for architects.

The post Two types of innovation appeared first on SogetiLabs.

The IT-product lifecycle

Every IT product has a lifecycle from cradle to grave. To ensure creating business solutions that contribute to business success an integrated process of Product Lifecycle Management (PLM) is needed. To succeed, PLM must be frontloaded with quality measures instead of checking quality at the end and wasting effort, money and precious time on fixing defects. To achieve this the organization needs a new and integrated way of collaboration between people using relevant, adequate and integrated tools to minimize efforts and work towards implementing the business solution in a way that gives confidence that the business value will be achieved, both on the short and long term.

This goes for traditional digital systems but just as well for systems that use artificial intelligence, robotics or any other new technology. Managing the activities during this lifecycle often is very challenging. The product lifecycle spans a very long time in the case of digital manufacturing machines.

Figure 1 The product lifecycle from cradle to grave.

Cradle

The process to create a new digital system often has many different activities that may be performed sequentially or in parallel. Amongst these activities there need to be quality assurance activities such as testing. Very important to bring the development up to speed is to have early quality measures, such as reviewing requirements, designs, models and/or user stories.

Use

Systems in the manufacturing industry may have a very long maintenance period during which changes must be monitored. In Product Lifecycle Management, many different systems are used to record all relevant data about the development process. The connections and exchange of all data is often a challenging process for which the use of tooling is inevitable. The use of AI-powered tools is a very promising way of improving this monitoring of the process as well as an improvement of the efficiency of the quality assurance and the product lifecycle activities as a whole.

AI-powered dashboards enable a continuous monitoring of the system during its lifecycle and can show when activities have an effect that goes beyond previously set thresholds or tolerances.

Grave

Every digital system at one point in time will be retired. Although switching off (parts of) a digital system may seem like a trivial action, in reality retiring a system is a complex process. Digital systems have an Information and operational part. Also called the IT and OT. Switching of an OT part (hardware) impacts the IT part (software). For example, the IT part continues with new OT systems.

Retiring hardware also includes making sure there’s a new system, converting the relevant content to the new system but also making sure the history of the system is preserved. Migration is the term often used in this context.

AI has good use in migration projects. AI is helpful for disclosing historic data, especially when it is stored in raw format. This is also known as digital archeology.

Digital archeology and testing

With the increasing speed of evolution of information technology, we also see an increase in the speed of retirement of products, systems and technologies. One aspect is storage technology. Due to legal regulations companies need to be able to keep their systems of record accessible for a certain period of time. The actual time differs per country and industry, but a period of a decade is very common. Back in the 1970s when all administrative work was done manually it was easy to keep an archive accessible. Just store the ledgers in a safe place and when demanded retrieve it and hand it over to the interested party such as the tax office. Nowadays, keeping an administrative system accessible for a decade is a very challenging task. Remember that it’s just a decade ago that the CD-ROM replaced the Diskette (or floppy disk) and already today it’s hard to find a computer that still has a CD-ROM drive. And another thing is the software. Companies regularly switch database systems or their ERP software. After such a system retires, they may still have the data, but they don’t have the software to interpret the data.

And this is where digital archeology comes in. It generally has two stages:

• Retrieve the data
• Interpret the data

For retrieving data old technology is often needed to actually read the data from an outdated storage medium. Recently we heard about an interesting case of using the tape drive of a computer museum to restore a website from the late 90s.

Once the data is available, artificial intelligence comes to the rescue. Using machine learning, especially supervised machine learning that brings in business knowledge, the dependencies and links in the data are determined.

This way the reconstruction of a retired system is enabled.

Before the retired system is fully reconstructed, we want to make sure the conversion of its data is correct. Digital archeology is the first stage in testing for these kinds of situations. Small amounts are converted and checked. In this case domain knowledge is of great use. Absence of domain knowledge is not an unfamiliar situation with older systems. In that case machine-learning algorithms look for correlations in RAW data and match it with known datasets. Machine learning is now used to providing us with potential outcomes of conversion. Human checks are still needed in order to see of the conversion makes sense or not. If OK, the next phase can start, and large-scale conversion takes place. Tests are executed after this stage as well to check if all conversion went well.

Product Lifecycle Management

During the product lifecycle, there are basically four interacting activities. They are shown (counter-clockwise) in the holistic view of Figure 2.

Figure 2 The four-segmented holistic view of creating and implementing business solutions.

Using this holistic model helps us to improve. The improvement has two focus areas:

• Manage and improve the product quality throughout the entire product lifecycle
• Improve the efficiency and effectiveness of the activities in the lifecycle process

An important principle from Lean product management is implemented to avoid waste in any kind. With a focus on the product risks and business risks, the right approach to product lifecycle management can be chosen. To achieve the desired level of quality, just enough efforts (time and budget) are spent in every stage of the lifecycle.

This approach to quality management enables the shift left move that enables early quality. With shift left we mean to move towards an earlier moment in the product life cycle with quality activities. In product lifecycle management, this is done by implementing various measures, methods and approaches. Testing, checking and test-driven development are examples of methods and approaches. An example of an important measure is the integration of all tools used to manage and monitor the product throughout the lifecycle.

Rik Marselis, management consultant digital assurance & testing

This article was based on the following books, they may serve you if you are looking for more depth:
– Testing in the digital age; AI makes the difference (2018) by Tom van de Ven, Rik Marselis & Humayun Shaukat
– The PointZERO vision; (2012) by Rik Marselis & Ewald Rodenrijs
– TPI NEXT; (2009) by Gerrit de Vries, Ben Visser, Bert Linker, Alexander van Ewijk, Loek Wilhelmus and Marcel van Oosterwijk

The post Product Lifecycle Management in the digital age appeared first on SogetiLabs.

January 2007, the iPhone was announced, ten months later Android debuted their first version along with an early version of the Google Play store. In another ten months the iPhone 3G would come bundled with the Apple App Store app. Within less than a year these two app stores would create an entirely new avenue for developers to easily distribute their own applications to users around the world. In order to maximize user retention, organizations need to create a unified experience across both app stores. The success of the project could be determined by which strategy they take.

Image Source

Choosing a Platform

Developing a mobile experience that will captivate users is the goal and the first decision to be made is -which platform to build your app with. The platform you choose could affect the cost of the project, the amount of time needed, and the limits of what the application is capable of doing. Today the four main platform categories are:

Native

• A separate application is built for each mobile operating system using native languages and tools
• No shared code
• Can require separate development teams for each operating system targeted
• Allows for complete control over the look and feel and access to any device features
• The Firefox app is written as a native iOS app

Cross-platform

• A single application is built using a single programming language
• Most code can be shared between each mobile operating system
• Able to access most if not all of the functionality that a native application can
• The Instagram and UPS apps are built using cross-platform technology

Hybrid

• Hybrid apps are built like a website, using HTML, CSS, and JavaScript
• Most code can be shared between each mobile operating system
• Accessing native features requires writing code in the native language if an existing plugin is not already available which can be time-consuming
• The Facebook app was developed as a hybrid application until recently

Web

• Should have mobile-first designs, allowing them to look good on small devices and desktops
• Most code can be shared between each mobile operating system
• Allow for the largest reach of any other option since they can run on any existing device and any device in the future
• Lack the native look and feel that native apps have, do not have access to all device features and are not an option if offline access is a requirement

Image source

Companies today have a number of different channels that they use to reach consumers, a similar approach should be taken when designing a mobile app. It may make sense to use a different platform for each app depending on the requirements and the deadlines. Other considerations are:

Target Audience

Internal applications for a company’s employees may favor utility and quick turnaround over a rich user experience. At the other end of the spectrum a company’s main user-facing mobile app may require a very native look and feel with complex animations and access to native device hardware. If the emphasis is on utility and quickly getting an application into production, a web or hybrid approach will usually be the fastest option with cross-platform coming in third.

Budget and Timeline

Native applications will always be the most expensive and time-consuming option when targeting multiple operating systems, requiring one language or even one development team for your iOS application and another for your Android application. Cross-platform solutions can greatly reduce this cost by only requiring a single language for both iOS and Android applications. Web and hybrid apps are usually the fastest and least expensive options.

Development Team Skills

Your development teams existing skillset could play a role in deciding which platform to choose. If you have a team with strong JavaScript development skills, then a native application may not be the best option.

Image Source

Platforms of the Future

New platforms are always being created for the web and for mobile apps, promising better performance and supporting more platforms. The mobile app landscape will look drastically different 10 years from now. One new platform that could alter mobile development is Web Assembly. Web Assembly, also known as wasm, is a low-level language similar to assembly that can run in modern web browsers. Because it is a low-level language it can reach speeds similar to a native application. What makes this platform so promising is that it can interpret instructions from almost any language and run them on the web. This could turn cross-platform development on its head, instead of forcing a development team to write their apps using a single language, that team could create an application in any language they want. Web Assembly is still very new but is already being used in some extremely innovative ways and is something to watch for in the next few years.

Conclusion

A company’s mobile development strategy should be almost as varied as the number of channels it uses to reach consumers. An application’s requirements, its target audience, and the company’s development strengths should all be factors when deciding on a platform to use. Native applications can deliver the richest user experience but require the most time and money. Cross-platform applications save time and money while still achieving the same or a similar user experience to native. Hybrid apps are able to be developed in much less time and are great for situations where there are few native device functionality requirements. Finally, the web offers the greatest reach of any other option but has strict limits on accessing native device functionality and achieving a native look and feel.

The post Modern Mobile Experiences and Platforms appeared first on SogetiLabs.

As mentioned in previous posts a code generation tool is very efficient to generate repetitive code from a “model” and to boost the project’s starting.

Here’s a short video demonstrating how to use Telosys code generator ( http://www.telosys.org/ ) in its “Command Line” version in association with Atom text editor to generate a web application written in Python.

In this demo, the application is generated from scratch using a “DSL model” to define the project’s entities.

It covers the following phases:

1 – Project initialization

2 – Model creation (to define the project entities)

3 – Templates selection and download

4 – Code generation

5 – Launching the generated application

Demo’s technical environment:

• Language: Python
• Frameworks: Bottle and SQLAlchemy
• Database: SQLite
• Code generator: Telosys
• Text editor: Atom (with the Telosys extension)

The post How to generate a web application in 10 minutes appeared first on SogetiLabs.

We know we can’t trust bots.

What can we do to trust them?

Can other bots help us what that?

Continued from Part 2 – How Bots Learn, Unlearn, and Relearn

Test Driven Development

TDD (Test Driven Development) is a development method that moves the developers focus from the code to the design of the solution. 

TDD requires the developer to first write a test case as unit test, then write the code to pass the test case. Then another test, then another piece of code, then another test, etc. The test case can be analyzed without reading any code and therefore the developer can validate and improve the design of the solution, without thinking about the complex code. 

Whenever a test case is changed, the code will have to be changed, for the test case to pass. Therefore, it is “Test Driven” development.

Develop and govern bots with TDD

TDD is perfect for bot’s based on reinforcement learning. The developer will still create the required unit tests as KPI on the QA-layer, and the bot will configure itself, without the developer have to change any code. The developer will only have to add another sensor (input) or motor (output) and some KPI’s.

Darwin vs. Newton

Using TDD on bots would be a Darwinian approach to software development. TDD will improve the bots from version to version, by guiding the bots evolution with KPI’s. 

This contrasts with the traditional Newtonian development method (the waterfall method), where the universe is perceived as a mechanical clock: “if you know enough details, then you will be able to predict anything”. The Newtonian approach gives the ability to understand “how” things happen, while the Darwinian only give the ability to understand “if” things happen.

Let me give an example:

A ticket selling application, sometimes makes an error and sells tickets with a negative price value.

• If this application was developed with the Newtonian approach, then the source code can be debugged to find out “how” the ticket price became negative. This error can then be fixed.
• If this application was developed with a complex bot, then nobody would be able to understand and therefore debug the source code. A KPI could have been set to measure “if” the price would become negative, so it can correct the bot to not do that again. Nobody would understand “how” the negative price became negative, only “if” it became negative and “if” it has been handled.

The Newtonian method seems superior in the example above, but that is only for simple problems. Bug fixing in very complex IT-systems can become a nightmare. Solving a bug, might re-introduce an old one or create a new one.  Combining the Newtonian approach with TDD is defined as TDRL (Test Driven Reinforcement Learning). This won’t solve all problems, since a bot might do something unexpected, but a KPI can be placed, in order to prevent this from happening again. For example: If TDRL bot is used to find cancer in lungs from images, then a KPI could be set for the bot to delearn and relearn from the new data set. If the bot is not able to map the input and the outputs, then it could sound an alarm, that it requires more sensors (inputs) or human input to solve the problem. In a larger scale, the bot could request help from court or the government to solve problems in a democratic way. It would also create precedent for future problems.

In a way TDRL would be even more agile than agile is today!

If we trust the bots, then we might get the opportunity to understand them.

Any concept is either:

• understandable by humans or not understandable by humans
• true or not true.

A combination of “understandable” and “true” can give the following categories:

Image by Bartek / Bartlomiej Rohard Warszawski

• Fiction or fantasy: Understandable, but not true.
• Knowledge: Understandable and true.
• Magic: Not understandable, but still true.

Complex machine learning bots goes into the magic category. We can only know “the things the bots do” and not “how the bots do it”, and therefore can’t trust them.

Maybe one day we will get a bot (in a future version of Cognitive QA), that will be able to explain to us mortal humans “how the bots do it”  – and maybe the answer will be… 42.

Testing in the digital age is a new Sogeti book that brings an updated view on test engineering covering testing topics on robotics, AI, block chain and much more. A lot of people provided text, snippets and ideas for the chapters and this article (part 3 of 3) is my provision for the book, covering the chapter robotesting (with a few changes to optimize for web).

The book Testing in the digital age can be found at: https://www.ict-books.com/topics/digitaltesting-en-info

The post In Bots we Trust appeared first on SogetiLabs.

The volume of data continues to grow today within the Company. They concern all sectors of activity, all functional areas but also everyday life with the Internet of Things.

Companies are embarking on a battle to control large volumes of data using Big Data projects to store this data but what about their control, their quality and the purpose of this storage?

Business Intelligence (BI) has been around for more than 30 years for managing and using data to help decision-making within a business, with these 3 main phases:

– Data Recovery and Processing using ETL tools (Extract Transform Load)

– Storing Historic and Qualified Data via a Datawarehouse

– Restitution of these data according to different needs (Reporting, Analysis, Dashboarding, Story Telling)

Do BigData projects replace BI?

The race for data and its control today favors the construction of data-lakes, but does this data coming from several sources, types and heterogeneous qualities are intended to be just stored for their pecuniary value?

No, companies that will stand out will be the ones who will be able to use this mass of data to be able to make decisions by analyzing them. For this, Business Intelligence is crucial and its latest innovations allow to exploit this new data source which is Big Data. Those who make decisions are not Data Scientists or Data Analysts, but people in the business and managers who expect simple restitution of the data they need (Reporting, Analysis, Dashboard, Story Telling) and especially Qualified data.

Data-lakes are therefore a new source of data for BI that must be processed, qualified and historized so that these data can be used by business people to help them make decisions.

New trends in BI appear:

• Tools that can attack Data-Lakes as well as other types of data sources, such as Alteryx, Dataiku and Birst simplify the selection of data and their processing while ensuring that the data available to the trades are of interest and of quality.
• The trades today want more and more autonomy, BI self-service tools (eg Qlik Sense, Tableau Software or PowerBI) have just evolve the Ad-hoc Reporting to allow dynamic representations and more and more while simplifying use for the trades.

It is essential to keep in mind that the processing of data from all types of source, its qualification and its availability for trades remain complex tasks and requires expertise.

It is therefore essential for companies embarking on these BigData projects to think about the purpose of the data and therefore to plan BI projects at the same time, allowing them to make the most of BigData project data.

The post The Importance of Data and its Use: The Big Challenge Today! appeared first on SogetiLabs.