Previously at the Volcano Base I got a bit mystical with the Rite of the Magician. Since then, I’ve been enjoying Spring. It’s the perfect season to configure complex automated systems, and to support an agency with the AI parts of their pitch.

Stripe, as a company, has built something genuinely impressive. It has also built pricing pages that seems specifically designed to make you feel like you've wandered into a GCSE maths exam without revising.

I discovered this recently when recommending bank transfers to a client as part of an invoicing automation project. The financial case was solid. The time savings were real and measurable. The only problem was that nobody - including Stripe's own support team, who promised to call me back and then apparently fell into a dimensional rift - could tell us simply what it would actually cost.

My client, perfectly reasonably, started sending the kind of emails that consultants know well. Long ones. Exploratory ones. Emails that circle around a topic like a nervous cat around a hoover, raising concerns that were really just anxiety wearing the costume of objections.

For most of my career, my response to this would have followed a familiar script. More explanation. Perhaps a document. A spreadsheet or a mindmap, if I was feeling particularly motivated. A call to walk through it together.

All of these have something in common: they require me to do the understanding for the client, and then hand it to them already digested. I'm the one operating the controls. They're in the passenger seat.

That's not a great place for someone to be when they're trying to decide whether to trust something new.

Fifteen minutes with Claude Code and I had a working calculator. Drop in your invoice volume, drop in your average invoice value, see your monthly Stripe costs. No spreadsheet to explain, no caveats to walk through, no ambient sense that I was probably trying to sell them something.

I sent it over and waited.

What came back wasn't another long email about whether bank transfers were a good idea. It was a question about whether PO numbers could be edited on invoices.

Which is, if you're paying attention, a completely different kind of question. That's not "should we do this?" territory. That's "when we do this" territory.

The calculator didn't persuade my client. That's the interesting part. It gave them the tools to persuade themselves.

I have a reputation for making complex things simple, and I've always leaned on analogies to do it. But analogies are still me doing the work. A working tool hands the controls over. It gives someone agency over their own understanding, which turns out to be considerably more convincing than the cleverest explanation.

The conversation moved. Not because I pushed it forward, but because they could walk around the problem themselves and decide it wasn't as frightening as it looked.

I've been making things digitally for a long time. That's always been part of how I work. But making something used to have a cost - time, tooling, the overhead of turning an idea into something functional. What's changed isn't the instinct to build, it's that the production cost has collapsed to almost nothing. The craft now is in knowing what to build and when, not in the building itself.

If you're explaining something to a client, a customer, a colleague, or a nervous board member, consider whether a small interactive tool would let them reach their own conclusion faster than your most elegant explanation. It doesn't have to be complicated. It just has to be theirs to play with.

What's the thing you're currently trying to explain to someone that might work better as something they could just... try?

Some interesting stuff I discovered on my adventures.

Claude now has a voice - and it's surprisingly useful

Anthropic has rolled out voice mode for Claude on web and mobile, allowing full two-way spoken conversations rather than just dictation. Worth knowing about if you do a lot of thinking out loud, or just want to interrogate your AI assistant while doing the washing up.

AI isn't your new colleague - it's a powered exoskeleton

A product engineer makes the case that framing AI as an autonomous agent sets you up for disappointment, while treating it as an amplifier of your existing capability produces better results. Worth reading alongside this week's Mission Briefing.

The AI ethics boycott that accidentally became a marketing campaign

A growing movement is urging people to quit ChatGPT over OpenAI's political donations and military contracts - and recommending Claude as the ethical alternative. Worth noting that Anthropic has its own $200m Pentagon contract and was facing pressure to drop its guardrails to keep it. Nobody's hands are entirely clean in this industry. Worth knowing before you pick a side.