The Process · Four to eight weeks · AI-Native or AI Adoption
Name. Systematise. Automate.
In that order, without exception.
An honest, week-by-week account of what an AI transformation looks like from both sides - what the studio does, what your organisation does, and what lands by when. If you engage adnah, this is the shape of the four to eight weeks, occasionally up to ten for a larger or more constrained organisation.
4–8
Weeks, on average
03
Phases, ordered
02
Tracks - AI-Native or AI Adoption
00
Ongoing lock-in
The whole engagement, one bar
Phase 01 · Weeks 1–2
Name
Nothing is systematised until it is named. Nothing is safely automated until it is systematised. The order is fixed, whichever track you're on.
Track note
Both tracks start identically here - this is where we tell you which one applies.
We start with read-only access to your case system, drive, calendar and billing - or the in-house/government equivalent. Nothing changes for the organisation in the first two weeks. What we're doing is a structural audit: what you use now, how you actually use it, and what best practices you're already following, in your own language. By the end of week two, you have a written map of your six MATTER primitives, a bottleneck brief, and a clear answer to which track applies - AI-Native Transformation or AI Adoption.
Week by week
Week 1
Structural audit
Read-only access, one working session with whoever runs the work, five closed matters read cover to cover. No prescriptions yet - just an honest picture of current tools, current process, current best practices.
Week 2
Primitive audit and track decision
Working session on the six primitives - what a matter is, where actions live, how deadlines surface, how work is reviewed. Gaps are circled, not judged. We tell you plainly whether you're structurally suited to full AI-Native rebuild or to AI Adoption within your existing constraints, and why.
The studio does
- ·Read your five most recent closed matters end to end.
- ·Interview whoever runs the work and one support staff member.
- ·Write the primitive map in your organisation's own language, not ours.
- ·Deliver the bottleneck brief and the track decision.
Your organisation does
- ·Grant read-only access to case system, drive, calendar, billing.
- ·One 90-minute working session with the people who run the work.
- ·Point at three matters you're proud of and two that went sideways.
- ·Read the map. Push back on anything that doesn't match reality.
Deliverables · End of phase 01
- Written primitive map
- One-page bottleneck brief
- AI-Native vs AI Adoption track decision
- Named list of top three gaps, ranked
Phase 02 · Weeks 3–5
Systematise
The work product of this phase is an organisation that would keep running cleanly even if we removed every AI tool tomorrow. That is the point, whichever track you're on.
Track note
Identical either way - the difference between the two tracks shows up next, not here.
Now the artefacts start landing, built to fit your track and your values. A matter template. An action library for the top matter types. A role chart for every named action - doer, reviewer, decider. A structured deadline model. An evidence-anchoring rule that applies equally to senior people and to anything a machine produces. This is the phase where 'we have systems' stops being aspirational and starts being a thing you could point a stranger at.
Week by week
Week 3
Matter template goes live
The first working matter container lands on your next live matter - not a mock, an actual matter. The old way and the new run in parallel for a week.
Week 4
Action library and role chart
Written sequences for your most common matter types, extracted from your people's expertise and pressure-tested against closed matters. Every named action gets an explicit doer and reviewer, written down.
Week 5
Time and evidence rules
Deadlines move onto the matter itself. The one-sentence evidence-anchoring rule ships and applies to humans first. Nothing is automated yet.
The studio does
- ·Draft, revise, and pressure-test all templates against real matters.
- ·Extract your people's tacit playbooks into written sequences.
- ·Weekly working session; async on everything else.
- ·Refuse to move to Phase 03 if the systematisation is thin.
Your organisation does
- ·Use the new matter template on every new matter from week three onward.
- ·Let senior people spend two hours a week talking us through their reasoning.
- ·Adopt the anchoring rule on human-drafted work first.
- ·Say when something doesn't fit your organisation - we will change it, not defend it.
Deliverables · End of phase 02
- Matter template in production
- Action library for top matter types
- Team role chart
- Structured deadline model
- Written evidence-anchoring rule
Phase 03 · Weeks 6–8, occasionally to 10
Automate
Only now do agents get deployed - on the actions we named, inside the roles we defined, against the evidence rules we set. Automation without the prior two phases is theatre.
Track note
AI-Native automates without constraint. AI Adoption automates inside existing sign-off chains - hence the extra fortnight.
We plug in software - our own MatterOS and LexOS where they fit, custom-built tools where there's a real gap neither covers, or market-available software where that's the more seamless and sustainable choice. Between two and four production agents deploy: an intake agent that opens the matter and populates the template, a drafting agent that runs the first pass on named actions in your action library inside the doer/reviewer chart, a research agent that checks the organisation's own stock before generating anything new. Every agent lives inside the systematisation of Phase 02 - that is what makes it safe, and that is what most organisations trying to 'add AI' skip. Larger or more constrained organisations on the AI Adoption track take the extra fortnight here, since automation has to slot into existing sign-off chains rather than a clean rebuild.
Week by week
Week 6
Software selection and intake agent
We confirm MatterOS/LexOS, a custom build, or market software for each gap. The intake agent opens a matter from a signed engagement letter or equivalent, populates the template, and ships to production this week.
Week 7
First drafting agent
A drafting agent deploys against one named action - usually first draft on your top matter type. It runs inside the role chart, with the evidence rule enforced. A human reviews every output, always.
Week 8
Research stock and handover
Your prior research becomes a searchable stock the drafting agent checks before generating anything new. Written runbook, agent configurations, the whole thing handed to the organisation. You can run this without us. That is the success case, not a failure of retention.
Weeks 9–10 (AI Adoption only)
Second agent, within constraints
For organisations on the AI Adoption track, a second agent deploys against a different named action inside the existing sign-off structure, with extra time built in for the approvals that structure requires.
What gets plugged in
MatterOS
adnah's own system for Matter, Actions, and Team - the matter container, action library, and role chart.
LexOS
adnah's own system for Evidence and Research - evidence anchoring and the compounding research stock.
Custom-built
Built from scratch where a real gap exists that neither MatterOS nor LexOS covers.
Market software
Existing tools kept in place where that's more seamless and sustainable than replacing them.
The studio does
- ·Confirm software - MatterOS, LexOS, custom-built, or market-available - for each gap.
- ·Build, configure, and tune two to four production agents.
- ·Run every agent through the role chart before it ships.
- ·Write the runbook you'd need to hire against your own operating model.
- ·Leave, cleanly, on schedule.
Your organisation does
- ·Review every agent output in the first weeks - this is how the model gets calibrated.
- ·Adopt the runbook.
- ·Decide whether a light monthly retainer makes sense.
- ·Publish your fixed prices, if you're ready and on the AI-Native track.
Deliverables · End of phase 03
- Intake agent in production
- One to two production drafting agents
- Searchable research stock
- Written runbook
- Clean handover
Before you apply
Not sure whether you're looking at four weeks or ten? The AI Readiness Audit scores your six primitives in two minutes and points at your likely track. For the structural reasoning behind why the track exists at all, read the thesis.
A limited number of transformations, each year
