How we write content
Most of the content on this site (web copy, insights articles, case studies) was written with AI assistance, and some of the diagrams and images that go with it are AI-generated too, from a spec we write first. Every idea in it came from us, and a person rewrote or approved every line, and reviewed every image, before it published.
The process is the same each time. We work through the structure with an AI assistant first, arguing about what the piece is actually about and what order it should go in. Then the assistant interviews the writer, section by section. It asks questions, the writer answers in a paragraph or two, and it drafts that section from those answers. The writer reads it, pushes back, and we settle before moving on.
When a full draft exists, the writer goes through it again from the top and rewrites whatever does not sound like us or does not meet the bar. Some sections survive that pass untouched. Others get taken apart.
When we cite an outside number, someone opens the source document and reads it before the number appears anywhere. A summary of the source does not count.
The ideas come out of the interview, which means they come from us. The assistant helps with structure and sentences. What you read here is what we think, and we will stand behind all of it.
How we work with clients
This varies by engagement, on purpose. A law firm and a landscaping company do not have the same constraints, and pretending otherwise would be a worse answer than the honest one.
We work to your AI policy. If you have one, it governs. If you do not have one, we will help you write one, and until then we default to the conservative reading: nothing sensitive goes anywhere it has not been cleared to go.
What does not vary is ownership. A person reviews everything that reaches you. The findings, the recommendations, and the judgment calls inside them are ours. If you want to know which parts of the work were AI-assisted, ask, and we will walk you through it.
How we build solutions
Writing code is the part people picture. It is the smaller half of what we do with AI.
We use AI to review as heavily as we use it to build: code review, security review, dependency and access checks, and a standing view of which reviews a project has gone too long without. AI makes that sweep cheap enough to run constantly. It still does not get the final word. A person reads the findings, decides which ones are real, and owns what happens next. Nothing gets called clean because a tool failed to flag it.
The same split applies earlier in the work. A person decides what to build and what to leave alone, and the planning before the first line matters more than the speed after it. We wrote about that approach in Context-First Development.
What we never do
Hold us to every line below.
- We do not publish fabricated data, statistics, or citations. If a number cannot be sourced to a real document we have opened, it does not appear. Where we cite something, we link it.
- We do not invent client results. Every result, quote, and testimonial on this site comes from real engagements and real people. Where figures are illustrative rather than measured, we say so on the page.
- Nothing reaches you unread by a human. No article, recommendation, or piece of client work goes out without a person reading it end to end.
- Client information does not go into unapproved tools. Confidential material only enters tools cleared for that specific engagement.
Why we publish this
Our tagline is "AI Assists. People Lead." This page is how we put that into practice.
There is also a practical reason. The businesses we work with are trying to decide how much to trust these tools, and the honest answer is complicated: enormously useful, genuinely fallible, and dependent on the process wrapped around them. We would rather demonstrate that with our own work than assert it in a sales conversation.
We recommend the same to the businesses we advise. If AI is part of how your company works, saying so plainly and in your own words costs very little and builds more trust than most marketing does.
If you have a question about any of this, or you think we have gotten something wrong, please let us know.