Workflow
You cannot automate a workflow you have not mapped.
37%
of teams that change the work start by fully owning one workflow, then scaling.
Deloitte AI Institute Pulse Check, 20261Teams describe the official process. The expensive work lives in the exceptions, the side spreadsheets, the Slack follow-ups, and the person who knows how to get a case unstuck.
How I see it
Map the work before you automate it
Discovery is the difference between automating the brochure and changing the work. It follows a completed case from request to result and writes down every handoff, queue, system, and judgment call.
The map should include the unofficial path. Where do people copy data between systems? Where does a senior person rework a junior packet? Where does the case wait because nobody owns the next step?
Once the path is visible, the economics become visible. Time sits in a few steps. Quality problems sit in a few judgments. Those are the places AI might help, and the places a simpler fix might be enough.
Discovery also decides the path. If the process is sound and two steps create the delay, improve those steps. If the process is a pile of historical workarounds, redesign it. If one outcome is valuable and multi-step, an agent may belong there.
How the engagement runs
What you actually buy.
A service is a sequence with an artifact at each step. It is not a transformation program you purchase as a bundle.
Map the unofficial path
Inbox, spreadsheet, chase, and the person who unsticks the case — not the SOP.
Cost per case and the real handoffs are visible.
A clean process map nobody recognizes.
Decide how far to change the work
Improve the current path, redesign it, or give one outcome to an agent.
The path matches the economics and the risk.
Automating leftover tasks on the old path.
Put people on the control points
Judgment, authority, and risk acceptance stay human. The rest can move.
Named steps are retired after the new path holds.
Every step still needs a person, so nothing got cheaper.
Do not buy this for
These engagements fail for predictable reasons.
Interviewing only managers
Managers know the intended process. The people who close the cases know where the time goes.
Mapping systems instead of cases
An architecture diagram is not a workflow. Follow one case until it is done.
Stopping at a swimlane
A pretty diagram that has no times, volumes, or error rates cannot support an investment decision.
Before we start
Questions that decide whether to engage.
Can you walk one recent case from request to completed result?
If the story changes by person, you do not have a process yet.Where does the case wait, and who is waiting on whom?
Wait time is often larger than work time.Which steps exist only because a previous system could not do the job?
Those steps are candidates to retire, not to automate.What does a good completed case look like, in writing?
If quality is tribal knowledge, AI will copy the inconsistency.What happens when the case is unusual?
If unusual is common, the happy path is not the real workflow.
How we size it
Cost of a completed casework time + wait time + rework time, valued at loaded cost, plus error cost when the case is wrong
Discovery should produce this number, even as a range. Without it, automation is a guess.
This is the wrong engagement if
Do not start this engagement yet.
- Nobody who does the work will be available.
- The company wants a tool recommendation in the first meeting.
- The process is already being redesigned for a reason that has nothing to do with AI.
- There is no completed-case definition, only activity.
A useful next step
Bring one expensive workflow. Get guided into production.
We guide the implementation, go deep on the technical path, and stay hands-on through operations — or tell you when a simpler answer is better.
Discuss an AI opportunity