With all the furore about Anthropic’s Fable, somewhere between Nuneaton and Rugby, I decided to see what would happen if I used it to build a complete collections system.

Now, this had to have login controls, role based access controls, be AI native, have case management, vulnerability and full treatment plans… KPIs, and seeded with 500 cases.

With a bit of care, some ‘quality’ prompting (and re-prompting!)… about 2 hours later, I had something.

Collect-RO-AR : Intelligent arrears management for modern lenders

At least on the surface level it looks surprisingly like a working collections platform. Not a slide. Not a wireframe. Not a vague prototype, but a demo system, and I think presentable.

Take a look for yourself, have a try at the link above. It did not feel too bad for an afternoon’s work!

Now back in the office and taking a longer look, I do need to stress that this is not a production ready system… there are clearly some gaps that need to be closed (the professional in me has to stress!)

• Full identification and verification
• Proper strategy management
• Detailed KPI reporting
• Operational logs
• Decisioning and controls
• Robust vulnerability treatment rules
• Financial difficulty segmentation
• Integration into a wider collections and recoveries environments
• Change controls and governance
• Supply chain controls
• Testing and assurance…. and more

However, if I can create this, this quickly and I am not a software developer or coder, imagine what someone who really knows what they are doing could create… and what would it look like after say, 6 months of work and development. It really makes you stop and think.

It is both exciting and slightly worrying.

Exciting, because it feels the speed of change is about to accelerate again… no longer are we being limited by the ‘HOW’ to do something… AI can now help. Going forward we are going to face a different limiting factor… the “What”. What should this look like? What is the process… What is the idea… this of course is where we need the human imagination (and often the more difficult question).

It is also a little worrying, as with this great power also comes great responsibility.

AI-generated software can now create something that looks very presentable very quickly. It can create an impression of completeness and polish almost instantly. Dashboards, buttons, workflows and case screens all feel credible…. if you are not doing this you may be left behind… and it is very easy to make something look and feel like a real operating platform.

But looking the part and being fit for purpose are not always the same thing, and this also matters… a lot.

You would never expose such a system, not fit for purpose, to real customer data… potential data security issues, functionality or process logic errors, big questions around controls and future support all could be hidden away underneath.

My collections system was created quickly from a couple of prompts, for example… was there an in-depth understanding of the code, what if I need to make changes, nope… How would you know for similar systems if there are any vulnerabilities in the code.

Also for every time those ‘[AI] need to do this on your computer, do you approve’ boxes appear on your screen, do you really research and understand what you are accepting…. I thought not…!

And, this is before we get to supply chain controls. Do we really know who is behind, companies in our processes… for AI, in this new API landscape?

Of course, much of this does not matter for a demo or ‘toy model’, such as the above… but in production it is critical. This technology is undoubtedly incredible and my prediction is we are on the cusp of a new wave of innovation in software, functionality, also in price.

However, whilst we should adopt and embrace it, we also need to make sure we do not lose our heads in the excitement.

• Engaging experts to review and assess new technology – ensure it is robust, meeting process and purpose
• Fully understanding process and technology supply chains – data design and controls
• Implementing robust governance, including AI policy and an understanding of change controls

The world is about to drop a gear to accelerate and we need to be able to keep up. An interactive demo may look great, but we need to understand ourselves and question what is the thinking underneath.

… the new age is dawning… enjoy the rest of the week.

This is also one of the topics we will be discussing at our next AI meeting on 15 October 2026 in Milton Keynes. Join us to find out how firms should think about the risks, controls and opportunities with AI. Please come along. Click the link below to register and join us.
AI Implementation Forum: Governance, Standards and Practical Implementation

With the World Cup well underway and temperature in the UK rising this week all I can think of is hydration breaks. No I am not talking about the hydration break you may have in the bar, when your team wins!… no I am talking about those ‘hydration breaks’ that have been happening throughout the matches for the teams to sip some water, whilst mainly marvelling at the strength of modern air conditioning.

Now my meetings are typically around an hour, so longer than the 45min in a half, yet despite it being warmer in my office than on most of the pitches (roof closed), I don’t seem to be getting those hydration breaks myself… I have to wait for the next hourly break. And, by looking at what has happened when you do, you can can see why this is.

Every time the match pauses, even for a quick glass of water the game changes. It may have been in mid flow, building in excitement, getting ready to score, yet it stops and the break seemingly resets the teams… and once it stops it takes a while to get back up to speed.

We would see exactly the same pattern back in the office. In a meeting, engaging discussion, ideas bouncing around and you are making really great progress…. but then the phone rings, someone or something interrupts you and everything stops. They you go, hurtling back to stumbling around trying to find momentum again… the heat has left the room, it has broken the flow.

It is a bit old fashioned I know… but it is important to make a cup of tea between meetings not during!

At its essence this is about flow… when in flow, time seems to fly, things get done, stuff gets shipped… once out of flow, everything is harder and getting back… that is an even harder struggle, often you have to restart the whole process again.

In our multi-tasking world it feels this is becoming ever more common… notifications, emails, WhatsApp messages, news headlines, even world cup scores they are all coming at us think and fast. Everything is trying to grab our attention, to distracting us from what we are doing, and get us to do what they want… now.

… And, with AI tools I am starting to find this making things even worse…. everything these days seems to need some micro-attention just to keep things ticking along… net result… progress slows.

Now I would never argue to not have a break for a cup of tea and a biscuit, but just like in the football, it feels like it is something we need to go back to deliberately planning for, so we have some uninterrupted time to play the game. Breaks that genuinely refresh… not manufactured or imposed ones, so something else can get done.

This is of course the same in the office, off and on the pitch… we will all be more focused and feel more rewarded as a result or our efforts.

So as the temperature rises this week… something to kind in mind… and not too much “hydration” during the matches this week….. have a great week, everyone.

Last week I reached AI overload. All the new features and that multi-threaded, multi-tasking feeling of just trying to keep up with it all.

So from this last weekend it has quite deliberately been an AI detox week. Just time in the garden, largely pulling weeds out of the lawn and then back in the office with spreadsheets and emails… ahh the good old days.

Stepping away to get a bit of downtime I think it has worked… and has given me a bit space to notice something…

Is our relationship with AI changing?

For much of the last year, AI has been a great support, and a very useful tool for sure.  With tasks loaded to complete, AI has become my handy helper… but it was largely still me running the task, actually doing the work.

This was not how it felt the last couple of weeks.

Development is now running at such a pace, and with my new set up in the office, that today it really feels much more like I am AI to do the work now… then reviewing the output, catching mistakes, refining prompts and re-directing.

No longer having to get it all done myself, being able to manage and organise, to have it done for me… I can now delegate. It is a feeling of getting promoted.

But, the thing is, this is also a different skill set… for those that have been managers, something they will recognise… and of course, as with real life, once you are managing things, everything become more complex… make sure you explain things clearly and accept that things are not always done exactly the way you would do them yourself!

I am wondering if this is happening here too. Are these also skills we are going to need going forward… AI may not be just about the technical, coding, integration with APIs… but also how to I effectively manage a bunch of agents, to deliver high quality output (to spec), to be effective and efficient… Management skills.

Are you keeping up?

And then there is another problem – managing AI energy. No, not energy for AI… but my own.

AI has this incredible capacity to get things done. It doesn’t have a Thursday afternoon slump, it doesn’t have a hankering for an early exit and curry on a Friday night and it certainly doesn’t need 2 cups of tea, a coffee and an undisclosed number of digestives to get going in the morning!

It is quite literally, the keenest employee you will ever have.  It sounds great… if you can keep up. By Friday I was exhausted and reminding myself I was not a machine!

So in this human and machine ecosystems we are developing is this something else we need to consider… and this may not be a phase, but the start of a new type of relationship going forward.

AI, after all is probably never going back to being a fancy calculator. The way things are going we are going to need to have clear boundaries over what happens, who does what when… just like real life… it is us the humans that need to create and enforced these rules.

It may just be the biggest irony that the more we use AI, the more we may need very human management skills… and not just for the machines, but also for ourselves too

Have a great upcoming weekend, everyone.