How it works
What happens after you say yes.
Six phases from first conversation to steady state, what each one produces, and what we need from you to run it. Most of the work is ours. The parts that are yours are named here so nobody is surprised by them in week three.
The six phases
From first conversation to steady state
AI readiness review
We find out where AI is already being used, including the accounts nobody told you about, what data exists, what of it is sensitive, and what your industry or your insurer requires of you.
Governance and budget plan
The rules and the numbers, agreed before anything is switched on: permissions by role, approved use cases, the seat plan, the usage allowance each seat carries, and the ceiling you want it capped at.
Workspace build
We stand up your tenant, connect single sign-on, configure roles, and bring in the documents and procedures your assistants will be grounded in.
First three use cases
We build assistants for three real jobs identified in discovery, in the departments that will feel it most. Working tools, not a demo environment.
Launch and role-based training
Live sessions by role, written playbooks for the work each team does, and champions named in each department before anyone is left alone with it.
Quarterly review
Usage, spend, and outcomes against what we agreed, plus the next set of use cases. This is the appointment that keeps adoption from decaying.
Phase one, in detail
The readiness review
This is the only thing you have to agree to in order to start, and you keep what it produces whether or not you go further. It takes a few conversations and some read-only access, not a project team.
- Where AI is already in use. Including the personal accounts nobody has told you about. This is usually the part that changes the conversation.
- What it is costing today. Subscriptions across expense reports, metered API bills, and anything renewing quietly.
- What data is in play. What people are pasting in, what is sensitive, and what your contracts or your insurer already require of you.
- Where the time is going. The recurring, high-volume work by department, which is where the first workflows come from. The industry pages show what that work looks like in six kinds of business.
- What it would take. A seat plan, the usage allowance and ceiling to go with it, and a governance outline you can take to whoever signs.
What we need from you
Your side of it, honestly stated
A managed service should not become a project you have to manage. But it cannot run on nothing, and the engagements that stall are always the ones where these were assumed rather than agreed.
One decision-maker
Somebody who can approve the governance rules and the budget without convening a committee. At most companies this is the owner, the CFO, or the COO.
A few hours per department
Discovery needs an hour with each team to find the work worth automating, and people need to attend their own training. That is the real time cost.
Identity access
Coordination with whoever runs your Microsoft or Google tenant to connect single sign-on. We work through your IT provider rather than around them.
The documents worth grounding on
Your handbook, templates, procedures, and prior good work. Assistants are only as useful as the material behind them, and generic ones get abandoned.
Then, every quarter
The appointment that keeps it alive
Adoption decays. Not dramatically. It just quietly stops being anyone’s priority. The quarterly review is the standing appointment that prevents it, and it is the single most important recurring thing in the engagement.
-
What the work actually was
The kinds of work that went through the workspace, by department and ranked. This is what tells you what to standardize and what to stop, and it is where the next build list comes from.
-
Spend against the allowance
What each team consumed versus what we planned for, and whether the ceiling is still set in the right place.
-
An honest seat recommendation
Which seats went unopened, and what to do about them, including when the answer is fewer seats. A managed service that only ever recommends growth is a sales channel.
-
The next three use cases
Taken from the top of that ranked list rather than guessed at, chosen with the departments, and built before the following review.
-
Governance revisited
New tools, new obligations, new roles. Rules set once and never revisited are how a policy becomes fiction.
-
What changed in the market
Models move fast. Part of what you are paying for is not having to track that yourself.
Find out what AI is already costing you.
The readiness review is where every engagement starts. Where AI is already being used, what data it is touching, what it costs across the expense reports, and what it would take to bring it under control. You get the findings whether or not you go further.