The method
Our method. How we decide what to build.
Five levels, from manual work to governed autonomy.
A five-level model that places any business on a path from manual work to autonomous, governed agents. We score where you are, and where the next defensible step is. Nobody should jump to the top. The model exists to stop you wasting money trying.
- L0
Manual
Work done entirely by people, no AI in the loop.
- L1
Assisted
AI used ad hoc by individuals, no system.
- L2
Augmented
AI embedded in specific workflows with human drivers.
- L3
Orchestrated
Agents coordinate multi-step work with human approval gates.
- L4
Autonomous
Governed agents run workflows end to end; humans supervise by exception.
From the stated problem to the real one.
A first-meeting instrument that descends from symptom to root, so we solve the real problem, not the stated one.
Presenting symptom
What you came in saying.
Value driver
The metric that actually moves money or mandate.
Binding constraint
The one thing truly holding it back.
Decision-maker utility
RootWhat the person signing off actually optimises for.
Four honest calls, not one default answer.
The diagnostic resolves every opportunity into one of four calls.
Build
Strong fit, clear value. We ship an agent.
Adopt first
The answer is an existing tool plus adoption work, not a custom build.
Seductive waste
Looks exciting, will not pay back. We say so.
Walk away
The wrong problem for AI. We tell you, and keep our credibility.
Most consultants only know how to say Build. The honest ones know all four.
Start here
See where your company actually sits.
Ascent runs the method on your own inputs, free, in about fifteen minutes. It scores where you sit, names the one workflow worth building first, and prices it in dirhams.