Artificial Intelligence – Creating the Environment for Adoption

Part II: Best practices and opportunities

In the previous post, we were describing the main blockers that are preventing businesses from a successful adoption of AI. We identified and analyzed these five areas: vision and commitment from executive teams and boards, organizational culture and change management, data readiness, talent, and, finally, standards and regulation. 

It is time now to focus on how to overcome these barriers, identifying best practices and opportunities from the experience of successful AI adopters. They are a great inspiration for companies looking to accelerate AI adoption within their organizations. 

Many of those early adopters have one point in common: they started their AI journey using an insight engine, as a way to run a trial and test AI benefits and capabilities. 

Relying on Artificial Intelligence (AI) and Natural Language Processing (NLP), an insight engine enables a fast get-started environment and provides companies with an out-of-the-box functionality to speed up the first deployments. 

Any organization thinking about implementing AI and more concretely an insight engine should be mindful of the following considerations to ensure a successful rollout:

  1. Decide how and where it should be deployed first

It is advisable to choose carefully which department or function will be the first to use and deploy an insight engine. This selection should be based on two main criteria:  feasibility and impact. 

If a department requires too much upfront work to get insight engine use off the ground, then it doesn’t make any sense to begin with that. Similarly, any new technology implementation needs quick wins to show tangible impact and value, and insight engines are no different. Getting this first implementation right paves the way for other functions to get on board, creating momentum that can build across the organization. 

Based on our experience implementing insight engines, the most common initial use cases are: 

  • Sales: Delivering a real-time understanding of customers and markets 
  • Marketing: Gaining market and brand insights from all your data 
  • Service: Powering the automated resolution of support and service tickets in real-time 
  • Risk: Keeping your business safe 
  1. Access to a demonstrator App 

Make use of a demonstrator app. This demo needs to deliver a thorough understanding of features and functionality and demonstrate the value an insight engine can bring to the organization. 

It is also essential at this stage for a clear view of the implementation method and a strong outline of what is required by the client to ensure success. 

  1. Address technical requirements upfront

There are many technical requirements for an insight engine implementation (too many to list in full detail here). But having a good understanding of these requirements is crucial for any organization that is even considering an implementation. How will the infrastructure be set up? Who is responsible for the data loading and enrichment? Is all the data available? Is there support for the supervised Machine Learning (ML) to help fit models? These and related questions need to be answered as early in the process as possible. 

  1. Get training and support for your staff

Any insight engine will only deliver the desired ROI if it is a technology adopted by all the target users, no matter if they are tech-savvy or not. Although the initial user group must be chosen carefully, insight engines are a technology suitable for any knowledge worker. Furthermore, they must be intuitive and easy to use. Employees must be given as much training and support as they need. 

  1. Connect Pilots to actual business ROI

Pilot projects are an essential part of any technology implementation. They make it easy for an organization to trial the technology on a much smaller scale and understand what benefit it can bring to the business. 

Pilots are most effective, however, when they are tied into Return on Investment (ROI). Senior decision-makers in any organization must know how an insight engine will create efficiencies, save money, and impact the bottom line. A pilot must clearly define the project goals and demonstrate how reaching these goals will translate into ROI from the insight engine.


AI and its implementation through an insight engine offers companies a rich set of opportunities to strengthen productivity and improve customer service. It also creates exciting new paths of innovation; for example, by realizing value in adjacent industries from companies’ data. 

But AI brings big changes to the way companies organize data and decision making. And it requires that companies attract and retain hard-to-find data scientists—and have the patience to train them up in specific industries. 

Finally, companies need to manage the ethics of algorithmic decision-making. All of this requires senior leadership and vision; boards and senior executives need to “upskill” to provide that clear direction.

Posted in Artifical Intelligence | Leave a comment

Artificial Intelligence – Creating the Environment for Adoption

Part I: The blockers

AI or Artificial Intelligence promises to unlock growth and raise productivity. While some companies started to adopt Artificial Intelligence successfully, many industries remain relatively untouched by AI. How to increase the adoption of AI is now a key element to extend the business benefits of this technology.

The AI opportunity – why some firms are embracing it faster than others

Although most companies are still a long way off from a widespread adoption of AI, the companies that are adopting it are already reaping rewards. A pre-pandemic McKinsey survey pointed to an uptick in revenue in the business areas where it was used—and 44 percent said it had reduced costs ().

Beyond this direct impact on the bottom line, companies that have successfully adopted AI report dramatic improvements in how they serve their customers and run their operations. While companies see the value of AI in optimizing processes, most of them, however, had not yet harnessed AI to drive disruptive innovation—for example, by creating new products or entering new markets. 

What holds business back from adopting AI 

At this point, the key question is: what holds companies back from greater use of AI? And, beyond that: what challenges do sophisticated adopters face when building on their existing leadership in AI? 

Five key areas come to my mind: vision and commitment from executive teams and boards, organizational culture and change management, data readiness, talent, and, finally, standards and regulation. 

The first two barriers show up in different ways at all stages of adoption and, indeed, we mainly see general organization change barriers arising when new ways of working are introduced. 

The other three barriers may depend on the adoption level and are more specific to AI. For example, gathering data into one place is a big challenge when a company first sets out. For intermediate adopters, securing the relevant talent is then a blocker. And for the more advanced and sophisticated adopters, apprehension of negative regulatory response is a barrier. 

Let’s have a look at each one of these blockers:

  1. Lack of vision and leadership from senior executives and boards

A clear vision is needed from senior executives and boards if companies are to get going with AI and their leadership is needed to stay the course. Yet in many companies there’s a lack of understanding of what AI is and where its value and opportunities lie.

  1. Cultural and change resistance 

Deploying AI is for many companies a huge challenge in organizational culture and change management. Companies unfamiliar with the value of AI will have lower buy-in and lackluster implementation efforts. To bridge the unfamiliarity gap, those companies look for external talent, which is often met by resistance in the company as these outsiders have little industry knowledge. 

  1. Data (un)readiness 

In most of the companies, data is distributed across many IT applications, is in a usable format or not correctly classified. This is in many cases a major inhibitor, especially in the early adoption stages. 

  1. Shortages of AI talent

The shortage of talent combining both business and technical understanding of AI is a significant drag on adoption. Companies need data scientists, AI experts, software engineers, and operators (often referred to as DevOps) to turn an AI pilot into a scalable, robust, quality assured operational business tool. 

  1. Regulatory and ethical apprehension 

In quite some cases, valid concerns about algorithms carry risks to the company’s reputation. There’s an opportunity here for widely accepted and clearly articulated standards to control or remove bias, track performance, ensure stability of machine-learning models, and improve the way algorithms’ decisions can be explained. 

These are the main blockers that, from our point of view, are preventing AI to be widely adopted in today’s businesses. In a second post we will put the focus on best practices coming from successful AI adopters, as an example of how to overcome barriers and ensure a smooth AI implementation. Stay tuned! 

Posted in Artifical Intelligence | Leave a comment

Tag 1 auf dem Weg zum Betritt zu einer reformierten EU

Sorry, this entry is only available in Deutsch.

Posted in Uncategorized | Comments Off on Tag 1 auf dem Weg zum Betritt zu einer reformierten EU

Einmal Staatsräson bitte

Ein Blick auf die Karte genügt: Die Schweiz ist mitten in Europa. Im Herzen quasi. Und trotzdem schafft es unser Bundesrat das Dossier Rahmenvertrag mit der EU seit über 7 Jahren zu verdaddeln.

Jetzt auf der Zielgerade soll Guy Parmelin am Freitag nach Brüssel reisen und das Dossier mit Ursula von der Leyen diskutieren. Welch Ironie, ein Exponent der Europa skeptischen SVP solls richten…

Können wir hier in der Schweiz bitte alle mal kurz innehalten und für einmal das Richtige tun? Die geographische Position alleine macht deutlich, dass wir uns mit der EU finden sollen und müssen. Für einmal ist der Begriff ‘alternativlos’ tatsächlich angebracht. Wir haben diese eine Geographie und keine andere.

Plus: Keiner der ach so lauten Skeptiker hat je eine realistische Alternative vorgelegt. Dieser Stuss von Global Switzerland und Unabhängigkeit und internationale Alternativen ist genau das: Stuss. Natürlich können wir mit anderen weiter entfernten Nationen Handel treiben. Aber die Nachbarn werden bleiben und sie werden uns immer näher bleiben als die anderen.

Und nur weil wir einen schönen Schrebergarten haben, heisst das noch lange nicht, dass die andern nicht auch einen schönen Schrebergarten haben. Jetzt müssen wir mit der Clubleitung des europäischen Schrebergartenvereins uns ins benehmen setzen. Daran führt kein Weg vorbei.

PS: Am besten wäre natürlich in der Clubleitung direkt mitzuwirken. Will jeder ambitionierte KMU Chef schliesslich auch im lokalen Gewerbeverein.

Posted in Uncategorized | Comments Off on Einmal Staatsräson bitte

Warum das BAG und unser Bundesrat so hilflos agieren

In normalen Zeiten ist die Kernaufgabe des BAT Risikoverminderung: Kampagne um Kampagne wird dem Volk erklärt es soll nicht rauchen (teert die Lunge), du sollst nicht trinken (macht die Leber kaputt), du sollst keine Cola trinken (all der Zucker…).

Sprich die BAG Mitarbeiter werden selber darauf trainiert Risiken abzuwehren. Nun wird eine so trainierte Mannschaft mit der grössten Krise der öffentlichen Gesundheit seit 100 Jahren konfrontiert.

Eigentlich bräuchte es jetzt Eigenschaften wie Sully Sullenberger sie bei der Notlandung auf dem Hudson an den Tag gelegt hat: Kühlen Kopf bewahren, die Situation kontinuierlich bewerten und überlegte Risiken (Keine Rückkehr nach La Guardia, Landen auf dem Hudson) eingehen.

Das Gegenteil von dem was in den Köpfen der BAG Mitarbeiter verankert ist. Kein Wunder verdaddelt das BAG Team die Pandemiebekämpfung.

Mit dem Bundesrat verhält es sich ähnlich. In unserem auf Konsens geeichten System wird meist nicht der brillanteste Kopf gewählt, sondern derjenige der auf der Gegenseite in seiner bisherigen Karriere am wenigsten Leute vor den Kopf gestossen hat. Berset ist ein idealtypischer Vertreter. Ewiges Studium plus Doktorat, anschliessend direkt und ausschliesslich Politik, ausgeprägtes Ego, aber nie wirklich aufgefallen, und mit Bedacht keinen Bürgerlichen vor den Kopf gestossen, und sich damit wählbar gehalten.

In seiner Karriere musste er ebenso wenig wie seine tapsigen Untergegebenen im BAG nie eine echte Herausforderung meistern, nie wirklich harte Entscheide fällen. Kunststück, dass ein so geeichter Politiker nicht brilliert, wenn genau das gefordert ist: Umsichtig, weitsichtig ohne Rücksicht auf das eigene und fremder Egos Entscheide fällen.

Das Schweizer System profitiert in normalen Zeiten von diesem auf Risikoaversion und Durchwursteln aufgestellten System, es ist aber schlecht aufgestellt für eine Krisensituation.

Bedenklich ist, dass das Ego der Betroffenen ihn im Weg steht, um im entscheidenden Moment auf andere zugehen zu können weil sie sich der eigenen Grenzen bewusst sind.

Deswegen gehören sie letztlich abgewählt.

Posted in Uncategorized | Comments Off on Warum das BAG und unser Bundesrat so hilflos agieren

The Insights and the Future of Information

A Sony Walkman was – some thirty-five years ago – a revolutionary piece of technology. Small, lightweight, and beautiful, it was music on the go. We made mixed tapes, gifted them to buddies and to our first romantic interests. Many of us still have stacks of tapes in the attic.

Most of us have since long switched to playlists on popular services like Spotify or Apple Music. Instead of buying a CD with 12 songs, of which consumer research already showed us 30 years ago that most buyers were interested in only 2 songs, you can create a playlist that matches your taste (or that of your BFF).

It is a profound shift in the way music is consumed. After long and hard years of transformation from the physical to a digital model, the music market, today, is worth more than ever. Instead of a physical asset, the asset is the insight into your music tastes: The playlist.

What if the same would hold true for data?

95% of data is never used beyond primary use. It is created in a blink of an eye, written to some hard disk, and sits there idly, never to be touched again. For public information, we use search engines such as Google to find the information. For business information, we need to open countless files across any number of systems and applications to find the document that we know is there… somewhere. Small wonder that most data is never used beyond its primary use. The fact is, it’s simply too complicated and too time-consuming to get to that information. In a world driven by ubiquitous data, this needs to change.

What if relevant data would find us?

Say, you want to know more about a client relationship. What about instead of combing through piles of often unorganized and unstructured data, the relevant insights come to you, right on time, automatically? You’d no longer be one step behind your customer’s experience. Instead, you’d be developing this customer relationship right alongside your client, in real-time.

For this future to be the new norm, we need to map a new course for information: Content, Relevance, Timing and Destination together form Insights; computational tokens, so to say. These tokens need to be enabled to find their way to the best possible destination – automatically. They need to become self-aware of the context they operate in, finding their way around the informational universe they populate in a self-propelled manner. On the way, they might collate with other tokens to form new types of Insights.

These insights need to be highly relevant to an end-user, which might be a human operator or a machine driven by software requiring this input to make its next smart move. These Insights tokens will become tradable opening up data silos much like playlists opened up the world to more music than ever before compared to the data silo known as the CD. This system will be an intelligent, self-aware software layer that will power any company in a not-so-distant future.

PS: These ideas drive the development of Squirro, and it’s likely the key reason why Gartner named us a Visionary in the recently released Magic Quadrant for Insights Engines beating such heavy-weights like Coveo, Microsoft, Google, Sinequa, etc. Get your complimentary copy of the Gartner report here.

Posted in Uncategorized | Comments Off on The Insights and the Future of Information

Indexing is not enough

We all use search. Every day. Most of us use one search engine only. Type a keyword or two, get a result list, look at the results – most of us never go beyond the first three results. If not satisfied, redo with another keyword combination. Most of us never use advanced search forms. Over the past 20 years, we’ve all been primed to this interaction pattern. 

Search traditionally works by crawling data, creating an index, providing for a search interface, typically a single field search bar. Trillions of person-hours have been invested by the leading contestants to get the relevance of the result set right. It’s the first three results that matter. 

Back in 2004-2009, when we operated with the largest homegrown search engine here in Switzerland, our key metrics were 1.23 keywords as input. That is not enough to spell “Bern Restaurant”. It’s either “Be Restaurant” – no we don’t want to be a restaurant … or “Bern Rest” – ok, the saying is that’s what Bernese people do the whole day anyhow (me being one of them). Today with better type ahead, this has grown to a query length of about two words. 

The core principle remains: You need to go to information. 

Done efficiently, it is a wonderful way to access information. Large search providers and more focused providers such as Enterprise Search vendors have added layers of sophistication to the approach (e.g. ingestion optimization, profiling, relevance ranking, and many more strategies). The basic approach, though, remains the same. 

In a world of too much information, this approach has its limits.

You need to know what you look for. Only if you can describe – to some level of precision – what you’re looking for, will the result list be somewhat meaningful for your request. While it is straightforward to use “that restaurant” for a search, it is less so for more complex situations. Say you join a new organization. By definition, you do not know what others, before you have written, say, about products. How do you start your search not even knowing the product terms?  Or you try to find, within a large organization, who knows what about a customer situation? Most organizations have disjunct information systems – multiple CRMs, service management tools, file shares, and more. How do you find the relevant information for a quick remedy?

A different approach is required to get to the next level: instead of you going to information, the right insights need to come to you at just about the moment you need them to get your work done. You turn information provision upside down. 

What is required for this to happen: you need computationally aware informational objects. Sure, the data item will not develop cognition about itself. But the construed concept of the informational object must express in computational terms the concept and the notion of the underlying information/data. 

In essence, the information/data needs to be expanded into a concept of an informational object with lots of (probabilistic) meta information to get some level of cognition of its content and meaning. Besides, the system needs a good understanding of your current informational needs, in other words, it needs a good profile of you. 

To extract such insights of information/data and you as a user, the recent advances in AI are the catalyst of this transformation. An essential element is to approach this transformation to an informational concept or object not as yet another relational schema but as a continuous probabilistic re-compute: Information changes, situational changes, user preference changes.

The first three Bernese restaurants we rendered back in the days were not, in absolute terms, the best. They were the ones we thought – well, as the ranking algorithm we put in place provided – were the most relevant to your search and matching your search profile.  Today we extend this concept and think of this like a continuously cooked bouillabaisse and from which the system picks a bowl when it’s time for lunch, automatically. 

This information revolution that is about to take place is well summed up in this quote from Anthony Mullen et al.:In the 15th century, Copernicus introduced the shift from an earth-centric to a sun-centric view of the solar system. There is a Copernican shift underway in how enterprises handle data. The approach shifts the emphasis from relational schemas as the center of the “representation” universe to concept and object models expressed across semantic and machine learning technologies.

Plus see what we do over at Squirro: Read Gartner’s Magic Quadrant for Insight Engines, where Squirro has been recognized as a Visionary.

Anthony Mullen, Magnus Revang, Stephen Emmott, Erick Brethenoux, Bern Elliot, Jessica Ekholm2021 Strategic Roadmap for Enterprise AI: Natural Language Architecture, Gartner, December 2020

Posted in Uncategorized | Comments Off on Indexing is not enough

The public debate that does not take place

We’ll get through the Pandemic – eventually. The vaccines are effective. It looks like we’ll get over here most vaccinated by end summer.

This would be the time by our political leaders to start a debate about the future beyond the pandemic plagued last two years.

Most of my friends and my family and me actually are privileged kids. We survived the largest public health crisis relatively unscathed. We were restricted for some time in our freedom of movement, but that was about it. I acknowledge this fact.

Many others are not so privileged. People in the hospitality business, in event organization, fitness centres, shops assistants, many in smaller companies lost their jobs, their livelihoods, their perspectives in life.

We as a society have an obligation to each other.

Given the economic devastation that Covid has caused it would now be the moment to start a debate about how to shape the post-Covid future, how to help to folks that for no mistake of theirs lost their livelihoods. We will need a Marshall type program to get them back on their feet. And yes that might include that we for say 2-3 years pay a percentage point or two more in taxes (and close some of the egregious tax loopholes awarded to corporations and wealthy individuals).

Why shall we do that? Out of pure self-interest.

Provide our children a great place to live. A society that is a great place to bring up your children and you can provider them a perspective and all of that in a safe environment is an equitable society. If that social contract is broken, you get broken states populated by desperate and vulnerable people. Just ask people in formerly prosperous places like Venezuela or closer to home people in destitute banlieues and suburbs in say Manchester, Berlin or even Geneva and Zurich.

Just, there’s no such discourse by our elected leaders. They will miss the beat, yet again.

PS: Irony of it all: Such a program would enjoy popular support and would be a vote winner.

Posted in Uncategorized | Comments Off on The public debate that does not take place

Missing the beat – again

Our lovely government has been inept at handling the largest public health crisis for a century. This is not a statement with hindsight but an expression of frustration for more than a year.

After having failed at setting up a comprehensive testing and tracing regime, our government bumbled the vaccine rollout: Lonza, based in Switzerland, proposed the government our own vaccination production.

While the ‘we don’t publish numbers on weekends, because also our employees need time off, too’ comment of the ministry of health is Asterix & Obelix level parody (‘It’s 5pm, tea time. We’ll be back to battle after that’) the rejection of the vaccination production kills people.

And that is just the latest misstep. With the next one already visible on the horizon: The social and economic upheaval is significant. And instead of starting a national conversation about healing, about rebuilding, about solidarity (yes that might imply higher taxes) you hear, well, Nothing…

But then, this is something most likely – and the proof points of the past 12 months support the claim – beyond their skillset. Most of the folks in positions of relevance starting with our government minsters have been politicians all their live or most of their lives.

In a country like ours a political career is still built primarily on risk avoidance (Any too extreme position will cost you the votes across the aisle to become minister… ). A crisis is pretty much the opposite. The nature of a crisis is confusion. It requires taking calculated and at times bigger risks to resolve it. Not an operational pattern any of our leaders is used to.

Posted in Uncategorized | Comments Off on Missing the beat – again

Attitudes matter

Excerpt from an email conversation with a colleague of mine. It started with a comment on this video: Squirro App Studio.

Teammate: This video is no longer current.

Me: Can you explain to me what is not current here (except a portion referring to DSS instead of AIS)?

Teammate: Hi Dorian, That’s what I mean. It refers to DSS so it should not be distributed. Is my assumption not correct? Thank you.

Me: Dude, in a perfect world sure. Now we live in an imperfect world: So yes your assumption is not correct (until we have redone the video, btw are you volunteering?). 

Look I react to this because most recently people left and right have been nagging about things and that goes on my nerves: Startups by definition are not perfect. We told all of you (Remember me saying in your hiring interview: Startups are f**ing hard. Are you in for that?). 

Tesla produced shitty cars for the better part of the first 10 years of their existence (don’t believe me? check their resell value….). But they were after something (understand mobility in terms of software not hardware) and see where they are today…

Same with us: We understand the digital world in terms of insights not data. On that road things have edges. Life with it or else. Or to say it with Mark Watney in the Martian ““I’m going to have to science the shit out of this.”

So I would have preferred a comment of the type: ‘Use video with note only: still references DSS, will need updating’. Bonus points for saying: @author: Contact me, happy to help for an updated version.

You know like I do for requests you have for me. Instead of me saying: “Do your job, deliver results.”, I am a ping away and e.g. over the weekend help you to expand the SFDC playbook to make you successful. 

Posted in Uncategorized | Comments Off on Attitudes matter