Writing

Notes on how we work, and why.

Short, specific, written by the founders. No newsletter, no gated content; there is an RSS feed.

13 Sept 2026 · 6 min
When do you outgrow Zapier, Make or n8n?
At three moments: when a flow makes a judgement only your business can make, when it can fail unnoticed, and when flows start depending on each other.
Argues for AI & automation
13 Sept 2026 · 6 min
Replace the software you run, or connect it?
Almost always connect. The tools are rarely the fault; the gaps between them are. Here is the test that tells you which case you are actually in.
Argues for Business systems
13 Sept 2026 · 6 min
Why do our reports disagree?
Almost never arithmetic. Three people answer correctly and get three different numbers, because they are answering three different questions.
Argues for Data & analytics
31 Aug 2026 · 2 min
A cost model is a deliverable
The first question about any production AI system should be the cost per request at volume. Most teams find out from the first full invoice.
Argues for AI & automation
31 Aug 2026 · 2 min
Why we publish the evals with the model
A model without its evaluation set is a demo. The eval set is the contract: what the model is for, what it is not, and how you would know it broke.
Argues for AI & automation
31 Aug 2026 · 2 min
The pyramid is the problem, not the people
Most failed engagements were staffed by capable people. They failed because what the proposal understood never reached the people who built it.
Argues for Technology strategy

Start

Write a paragraph. A founder replies within one working day.

A 30-minute call with both founders follows. No deck and no proposal template, just a first read of whether we are the right shape for the problem.