6–12 weeks

Tactics

We read how your organization works and where AI would take real load off it, and hand you the Roadmap. Then we build it into that process, on the systems you already run, with the controls to show it holds.

A video call where we decide where to start. No slides, no pitch.

Who it is for

The team that owns the workflows in question, and the people who run the work that will change. Tactics is where AI meets real work, so it is written for the people who answer for that work.

Strategy settles what AI is for. Tactics decides what to do first and puts it in place.

1–2 weeks

The Roadmap

We read your actual systems and workflows, then recommend the best place to start, including what not to build and what to wait for. It is costed, sequenced, and neutral on vendors. Where the first move is technical, we build a scoped proof of concept so you can see what it does with your data.

What the Roadmap contains

A summary: what we found, what we recommend, what it costs, and what happens if you do nothing.
Your current state: the tools you license, the AI use already working informally, and what has no controls.
The recommendation: the one best place to start, and why that one.
What not to build or buy yet, and why.
A proof of concept, when the first move is technical, built on your real data and labelled as a reference implementation, not a production system.
The plan: steps, timing and cost.
When to revisit it: the assumptions most likely to age, and a date.
What is out of scope, said plainly.

The Roadmap is yours to keep whether or not we work together again.

5–10 weeks

The build

We put AI into the workflow that matters, on your own platform, with the access, logging and audit trail to show it holds. By the end you can check every box on four lists. An unchecked box is a decision we named, not an oversight we missed.

Governance

Who can trigger the workflow, and the list is enforced. Every run is logged and you can query the log. A person checks the output wherever it touches something consequential.

Grounding

Answers come from your own material, and that material is kept current. Someone has checked a grounded answer against the source.

Context

The instructions that drive the workflow are saved and versioned. Someone other than the builder can explain them, and they are not tied to one vendor.

Testing

It has been run on deliberately hard inputs. Someone who did not build it has tried to break it. There is a plan to test again when the model changes.

Start free

Find where your work stands before you talk to anyone.

Which work should AI do first?

A few minutes, all clicks. No email needed to see where to start.