AI Hardware craze & the reality of passive services

Rabbit AI secured 30 million in Jan 2024. Are we experiencing the rebirth of 50-year-old experience paradigms, or are we going down the path of repeating failed design history?
As a CTO or CFO, you have set aside a few million for AI-based projects in 2023. There are a few things to look out for to de-risk this investment.

Interactive product paradigms have been tested a long time ago. Dieter Rams’ designs from the 60s that influenced Apple’s iPod experience are just one of many successes from over 70 years of innovation work.

Despite these tried and tested design insights, AI Companies made of former Apple employees are bringing back products that defy those insights, claiming that they will work despite or just being unaware of those learnings.

It is not the first time that technology companies have claimed they can ignore the reality of user behaviour and needs just because the technology is so good. And it won’t be the first time they fail. How can you, as an organisation, create AI solutions that will fit into your employees’ and customers’ reality?

Three considerations are what every CTO or CFO should look out for when discussing your AI or new tech budget:

1) Garbage-in-garbage-out: Context is king!

AI isn’t magic. Work with it for two weeks, and you will see how limited it is. It is powerful, just like APIs, block-chain and, in general, data, but just like them, once they collide with the reality of human beings and their behaviour, if their endpoints aren’t designed right, what goes in and out of the system can easily be of no value or even damage you and your company. It is one of the risks and efforts that is often missing from budgets and value propositions. It is one of the most significant missing pieces responsible for the low success rate of investments and transformations. McKinsey said over 70% of transformations fail this way, and 98% of startups are often biased towards technology and miss their market fit because they ignore the customer needs.

In a recent project, we needed to re-design the IT services across a whole portfolio of companies to centralise and be more cost-effective. A great piece of enterprise architecture built a brand new single-sign-in and data system for dozens of companies with thousands of employees. Data needed to be cleaned up, and new ways of working needed to be in place to use the system, not in a human way, but the way technologists had designed it. People refused to make time for it or didn’t have the expertise at a location to do so. Everything ground to a halt and the knock-on effect cost hundreds of millions. When you looked at the programme planning, there was no contextual assessment of where the solution was being deployed. The people aspect was missing and under-invested.

Or take this news item about the AI Pin created by former Apple designers. It is a complete absence of understanding the context of how people interact.

Einstein famously said: “If I were given one hour to save the planet, I would spend 59 minutes defining the problem and one minute resolving it.”

If we would budget that way, I am sure we would see much less investment failure. The good news is that it wouldn’t take 59/60 of the budget to do the discovery. Research and data have become so cheap that they shouldn’t cost you more than 2–5% of the budget. It’s not a bad investment, given that, in my experience, it de-risks 20–50% of your project through cost reduction and error reduction and minimises the cost of maintenance, training, and time-to-market. Some of this is evidenced by an independent Forrester study, which examined six businesses implementing the discovery and testing process. Even without that study, it is common knowledge that later changes are more costly, so why not reduce those costs?

Your AI investment will do well if you invest in those 59 minutes to ensure you are not pumping garbage into your pipes.

As a senior leader, there should not be any excuse to spend 2% of your budget and time on proper discovery. Yet, I have seen it too many times over the years, and AI is exponentially more able to damage your investment and organisation if you are not aware of the context within which you invest. This is not a tech challenge!

Budget for context discovery! 2% can de-risk 100+% because if things go wrong, they can backfire beyond the investment budget and cause long-lasting damage to the culture or brand of your company.

The Post Office case has likely started with senior leadership having no visibility of how their organisation functions.

2) Passive Services: Stop developing active automation!

Sometimes, I want to get a notification or a relevant status update. Check my deliveries or account spending or know if a meeting is still happening. I don’t need notifications once I have received a delivery. I don’t need my team to check in with me every day. I also don’t need arbitrary traffic light statuses. Notifications have, for ages, been shown to be a nuisance at best. And they are often not personalised enough to be of value. I am still waiting for the email app that can filter out the 80% of spam notifications created by all my services. We need systems that ask for less of our engagement, not more.

This is why many website bots and ticket systems are so bad that people avoid them.

Probably three times this week, I had to deal with a bot, and I’m still thinking it would save me time. The nearly identical experience was:

  • The bot asked me my life story before asking what type of problem I had.
  • Then, he asked me to wait for a response from the bot or, if I was lucky, a helpdesk person dealing with likely a dozen requests simultaneously.
  • Gone are the times we sit on a phone for half an hour to get connected to relevant people. Present is when you skip to another tab in your browser to do some work and return to the bot 2 minutes later to a message that says: We’ve been waiting for you, but it seems you have left the conversation, so we cancelled it.
  • You do that about three times before you demand to talk to someone on the phone.

How this helps anyone on either side of the problem is anyone’s guess. The main issue is that the experience is based on active engagement and is out of sync with what a user expects. AI systems will likely be designed similarly if we are not careful.

The Rabbit has an example in its PR video of asking for a perfect travel itinerary. One sentence is asked, and it builds a whole experience.

It’s just that this is not how travel planning works.

  • We compare with partners; we move things around; we don’t just buy a package and be done with it.
  • Therefore, reading it out, etc, is just not good enough. Small tasks like book me x on Y at z will work; Anything more complex needs to be designed better.
  • The check-ins are essential. When can I let AI automate, and when should it check in with me?

A lot of what people need are passive services, but we keep developing solutions that are focused on active engagement. You also can’t automate and let the bot do the job unless you know how a person likes a job done. This means scenario development and research that no company wants to invest in today’s model.

However, it is where all the future value is. Active services are too commoditised. Doing more of those will not create as much value as AI hardware does.

As a senior leader, you should ask: Which metric will you improve to automate into a passive service, and is it the part that the end user wants to get automated?

If people can’t give you evidence of that need from users, do not sign off the budget!

3) The Trend: Transactional — Experiential — Passive

People pay extra for higher-quality services. People go for convenience and impulse buy when the context is right.

In 1999, Joe Pine stated that we live in an experience economy. Starbucks didn’t sell coffee; it sold the Italian experience and charged $4 for a coffee that McDonald’s charged $1.5 for. Getting a coffee fast was a commoditised service, just like buying freshly ground coffee and coffee beans.

Joe Pine — The Experience Economy (+passive)

1999 was a long time ago, and customer expectations have shifted significantly. Where did the new margins move to? Beyond experience, there are passive services. Whilst you enjoy a $6 cold-pressed coffee, your bots are collectively re-mortgaging your house and talking to other bots to bargain for a better gas supplier.

Look at Netflix or YouTube recommendations on how Uber balances that enough cars are on the road to pick up customers. These are early passive service examples, many poorly designed or in their early days. You don’t see them, so you don’t realise they create exceptional value and make companies grow and function.

There is a striking similarity with the value of good design. You will not see that the height of a stair is perfectly designed so that when you go up and down a staircase, it suits your body perfectly. But people like Le Corbusier spent a fantastic body of work perfecting it.

Most companies still struggle to design coherent experiences, but customers expect passive services already.

One of the biggest challenges of passive services is governance vs automation. When do you let the bot go ahead, and when does the bot need to check in with you? I will stop the flow and time-savings if I need to give an OK for every step. If I can reduce every employee’s time to sift through information from 1 hour to 1 minute, and I have 100,000 employees, I am saving over 12,000 working days.

Processes being complex and the human component being exponentially more complex, understanding context and reality is what creates the value for a passive service.

In about a year or so, no one will care about if a tool has AI in it or not. However, they will pay good money if your service helps them do complex things without doing Anything.

So, if you spend money exploring AI, forget hardware or outdated experience gimmicks. Focus on passive services. In any proposition, it should be clear how AI parts fit into a process and where the check-ins are with the bot.

In most internal processes or customer services, one step of AI implementation will not make the difference anyone will claim. To create value, you must have time and budget for scenario exploration. Scenarios are the processes that do not go perfectly because people’s needs aren’t linear, not everyone understands your jargon, and people have different mind models.


  • Know your context; know reality. A system is neutral. Expose it to bad stuff, and it will create bad stuff. Help people fuel it with value, and you will get value. Ask how the system will be driven by value. If that path is not clear, the project will be a mis-investment. If you can’t get your employee data into your talent solution or have it updated frequently, the system will be useless and, at times, will waste time and money.
  • Ask what human behaviour exists to ensure the correct data can fuel the solution. If training or a whole process change is needed, that’s a flag to put a budget against.
  • Get evidence of what size of passive service is created. Digitising alone often creates no extra value and can lead to costly transformation and change costs, nearly always leading to new problems that need fixing. Cheaper systems, in particular, often fall short and need fixes right after the initial go-live. The AI Pin is an excellent example of useless features in context.
  • Ask for the metrics that create the value. Even without more profound research and testing, the problem size the solution will fix should be known within reason. No solution fixes everything.
  • What is the alternative scenario? AI is a living system. People who use it are as well. Both sides will evolve, and most projects do not account for iterative costs vs iterative profits. This is what we call wicked problems, and they need new approaches and teams that can evolve values differently. Ask for how both value and investment will evolve.

AI in proprietary hardware is not the wisest investment. Passive evolving services are. This is where the value is and where our attention should be, understanding reality and behaviours.


Marcus Kirsch has over 20 years of experience in end-to-end product and service and process innovation and delivery on a single and portfolio level.

He is a keynote speaker and author of ‘The Wicked Company’, a book about how organisations can better solve wicked problems.


From ‘The Wicked Company’:

“ Marcus Kirsch’s superpower is to see problems at both the macro and micro scales. The Wicked Company surveys the landscape to help us understand how, when and why change happens; then he dives deep to help us figure out how to capitalise on it.”


You can contact him on LinkedIn or follow ‘The Wicked Podcast’ on Apple, Youtube and Spotify.

His second book will be published later in 2024.

Marcus and The WIcked Company are available for workshops and projects.


Forrester Report:
Dieter Rams: Book — Amazon Associate Link
The Wicked Company: Book — Amazon Associate Link
The Experience Economy: Book — Amazon Associate Link

Similiar Articles


Sign up today for monthly newsletters containing:

  • News and insights from your industry
  • Relevant thought leadership articles
  • Engaging video content
  • Notifications of our upcoming events
  • Networking opportunities with C-Suite leaders