Answer
Do you use AI in every tool?
No. AI goes into a tool only where it earns its place; it is a way of delivering the tool, not the product. Many of the most useful workflow tools need little or none of it.
Where AI earns its place
The job is to fix the problem you have so the crew uses the fix. Whether AI is part of that is decided in the Spec, based on the workflow, not on a wish to use it. Plenty of broken workflows come down to a handoff that nobody owns, a queue nobody can see or work that gets typed twice. Those are often fixed with a well-shaped tool and a clean connection between systems, with little or no AI involved.
When AI does go into a tool, the models and providers used are listed in the handover pack, and client data reaches them only through our own platform, using providers named in your contract. Client data is not used to train models.
What we decline
At the fit call we say no to a request to use AI everywhere without a specific workflow. At the Spec we decline tools that would take actions outside the company on their own, control of safety-critical equipment, money that moves without a person approving it, employment decisions, and binding commitments to customers without a human approval step.
Those limits are set before the build, not discovered at go-live. More on the method is on how we work.
Questions people ask
Will you take on a request to add AI across the company?
No. A request to use AI everywhere, without a specific workflow, is one of the things we decline at the fit call.
Is our data used to train AI models?
No. Where client data reaches an AI model, it does so only through our own platform, using providers named in the client's contract.
Will we know which models a tool uses?
Yes. The handover pack lists the models and providers used.