How an engagement works
The same path every time, and every step ends with your decision.
You always know what happens next, what you keep, and what's yours to decide. Nothing goes live without your sign-off.
01
Talk
30 min · free
02
Diagnose
2 weeks
03
Fix the foundations
1–3 weeks · only if needed
04
Build with controls
3–6 weeks per workflow
05
Keep it accurate
Monthly
If your foundations are already strong, we skip straight from Diagnose to Build. If they aren't, fixing them first is what makes the automation keep working.
Step by step
What happens, what you keep, and the decision that's yours at each step.
- 1
Talk
30 minutes · free
What happens
- ✓We look at where your team's hours go and what's written down.
- ✓We decide together whether AI is the right fix yet.
You keep
An honest answer. If AI isn't the right fix yet, I'll tell you what to do first.
Your decision
Whether to start with a Diagnostic.
- 2
Diagnose
2 weeks · fixed fee
Method phase: Align
What happens
- ✓We agree on your goals and on what the AI must never touch.
- ✓I check which AI tools your team already uses on their own.
- ✓I score your five foundations: ownership, documented workflows, written SOPs, clean data, and systems that fit.
- ✓I record today's time, cost, and error rate for your top workflows.
- ✓I design the solution on tools you can run yourself and turn it into a 90-day plan.
You keep
A signed confidentiality agreement, a readiness scorecard, a gap list, a 90-day roadmap, and a written scope with a fixed fee.
Your decision
Go or no-go, and which step comes next.
- 3
Fix the foundations
1–3 weeks · fixed fee · only if needed
Method phase: Document
What happens
- ✓We pick the workflows and name an owner for each.
- ✓I record your team doing the work and collect the files they use.
- ✓I audit the data, and you approve the cleanup plan before anything changes.
- ✓I clean the data, map the workflows, and write the SOPs.
- ✓Your team reviews everything, and I walk you through the finished package.
You keep
Workflow maps, written SOPs, cleaned data with a log of every change, a data dictionary, before numbers, and a readiness score.
Your decision
Is the operation ready to automate? We score the foundations again to find out.
- 4
Build with controls
3–6 weeks per workflow · fixed fee
Method phase: Govern and Deploy
What happens
- ✓You approve the design before anything is built.
- ✓I build it switched off, on tools you already pay for.
- ✓It is tested against a set of your real past work.
- ✓It runs alongside your current process until it meets the agreed quality bar.
- ✓Before go-live we confirm who owns it, what the AI can do alone, how to pause it, and how we'll measure it.
You keep
One plain-English Build Summary, a runbook your team can follow, the test results, and a governance checklist.
Your decision
Your sign-off. Nothing is switched on until you give it.
- 5
Keep it accurate
Monthly retainer
Method phase: Optimize
What happens
- ✓Each month we review accuracy, exceptions, and cost against your before numbers.
- ✓Everything is re-tested after any change to your tools, prompts, or process.
- ✓A sample of the work that ran on its own gets a human check.
- ✓A task takes on more only when it earns it, and drops back if it slips.
- ✓One new workflow each quarter.
You keep
A monthly report showing how the AI is doing.
Your decision
Which tasks have earned more responsibility.
What stays true at every step
You stay in control
Everything is built switched off, and nothing goes live without your sign-off.
You own what we build
You own every document, whether or not you automate. You can stop after any step and keep everything.
Your data stays safe
I work from read-only access or exports where possible, confirm a backup exists before changing anything, and get your written approval before merging records.
No surprises on cost
Every project is a fixed fee, quoted in writing after a 30-minute call. You know the full cost before anything starts.
Want the full picture of how AI earns more responsibility over time? See the Earned Autonomy method →
Find out what's ready to automate and what needs fixing first.
A 30-minute operations call. We look at where your team's hours go, what's written down, and whether AI is the right fix yet. If it isn't, I'll tell you what to do first.