An Ideal Customer Profile walks into a bar...
”If you want to make an apple pie from scratch, you must first invent the universe.” Carl Sagan
Previously at the Volcano Base I’d been trying to make “fetch” a thing. I mean “synthetic computing”. Since then, I’ve been implementing some automated business systems for a couple of clients.
I’ve even been invited to pitch for a systems project. But I explained I don’t pitch for things. So I’ve been invited for a chat instead. Which I’m preparing for as though it’s a pitch. So, yeah, I think I fluffed that one.
Mission Briefing
An Ideal Customer Profile walks into a bar...
For months now, I’ve been trying to figure out who my Ideal Customer is. Not in an existential sense (although that has come up), but in the tidy, professional, allegedly useful sense. The kind of thing you're supposed to do when you run a business and want to sound like you've got a plan. The fabled Ideal Customer Profile or ICP.
So I started collecting them.
They came from government departments and e-commerce shops. Security consultants. Service designers. Venture capitalists. International FMCG people who are sometimes CPG people, depending on the prevailing winds. SaaS platforms. Research outfits. Creative studios. Frankly, it was beginning to feel like assembling a dinner party where nobody knows why they’re there but the wine’s already open.
They didn’t look the same. They didn’t act the same. But they all said the same thing, in slightly panicked tones:
"We’ve got goals. We’ve got tech. But the tech isn’t helping with the goals. The systems are creaky. The admin is breeding under the stairs. We’re dropping things. We’re busy but confused. We can’t find anything we already know. Help."
It was at this point I realised I was looking at the wrong ICP. Not Ideal Customer Profile, but Ideal Customer Problem.
This was a bit horrifying. You see, Profiles are easy. They're tidy little dossiers. You can build them with drop-down menus and database filters. Problems, on the other hand, are elusive creatures. They hide under Profiles and only emerge when you’ve earned their trust.
You can’t search for them. You can’t run ads to them. You can only discover them in the ancient, pre-tech way: by talking to people until they start saying “Oh thank god, yes, that.”
Which is thrilling, in the way that realising your parachute is also a picnic blanket can be thrilling.
So now I’m doing something strange. I’m running a consultancy that doesn’t target people, but patterns. I look for signs of friction, fatigue, and “why does this thing even exist.” I listen for the sigh between “we’ve always done it this way” and “please make it stop.”
And if you’ve solved this puzzle yourself… if you’ve found a clever way to find problems before the profile steps in wearing a name badge… I’d genuinely like to know, because <shrugs>.
Classified Intel
Some interesting stuff I discovered on my adventures.
OpenAI Introduces ChatGPT Agent: An Assistant with a Virtual Computer
ChatGPT now "thinks and acts" by combining tools like Operator and Deep Research to run tasks in real time, always asking for permission before acting.
Why it matters: This isn’t just a smarter chatbot. It’s an AI helper stepping into your workflow, handing off the grunt tasks while you steer.
Perplexity Launches Comet: An AI Browser That Behaves Like a Second Brain
Comet embeds Perplexity’s AI into a Chromium browser, summarising, autofilling, automating, while keeping your data local.
Why it matters: Browsing is mutating from passive scrolling to active collaboration. Tools like this hint at a future where your browser anticipates your needs.
Anthropic Rolls Out Claude for Financial Services: AI with Audit Trails
Claude now serves up due diligence, benchmarking, and modelling with built-in traceability and compliance controls.
Why it matters: Finally, a generative AI that’s not just chatty but compliant. Built for high-stakes industries where every insight needs a receipt.
Your data is not shared. Unsubscribe with 1 click.
Member discussion