Home / AI principles
[ Principles ]
What we will and won’t build
“Human-first AI” is on every consultancy’s homepage including ours. Here’s what we mean by it, in terms specific enough to hold us to.
1. A person decides
Our systems draft, sort, score, and prepare. Sending, paying, committing, and deciding stay with you. Where we do let something run unattended, we say so and explain why the consequences are contained.
One system we built scores every completed sales call and files coaching notes with no human in the production loop at all. That’s safe because nothing leaves the company, every score points to a timestamped quote, and the person receiving it was in the conversation. Full automation is a decision about consequences rather than a badge.
2. It sounds like you or it doesn’t ship
Generic AI phrasing in your outbound is worse than no outbound, because it tells your client you didn’t write it. We build the voice base first and test drafts against real examples of your writing before anything goes live.
3. No black boxes
You get documentation and training on everything we build. If you want to take it in-house, you can. That’s the standard we build to, and it’s why handover is a deliberate process rather than an afterthought.
4. We’ll tell you not to build it
Some processes should be fixed rather than automated. Some aren’t repetitive enough to be worth it. Some need a developer beyond our technical scope. We say so, and we refer out when it’s outside our lane.
5. Your data is yours
Custom deliverables built for you are yours on full payment. We keep our own pre-existing tooling and methods, and you get a license to whatever of it sits inside your build. We don’t train models on your data, we don’t use your material in another client’s build, and we don’t name you as a client without asking.
Questions about any of this?
Ask them on a call before you spend anything.