AI Transformation Consulting
Map the Human Workflow Before You Automate It
A process diagram can look complete while missing the person who makes it work. Before proposing automation, trace the human decisions around the task, especially the help people seek when the normal process breaks down.
Illustrative manufacturing example
The task is visible. The judgment may not be.
Sales captures
A quote request
Engineering checks
The specification
Planning confirms
A delivery commitment
| Decision | Evidence needed | Human dependency |
|---|---|---|
| What did the customer mean? | Confirmed requirements | Sales checks ambiguity with the customer |
| Is a substitute suitable? | Specification and constraints | Engineering reviews an exception |
| Can we promise that date? | Capacity and production implications | A planner consults an experienced operator |
A teaching example, not a deployed customer workflow.
Follow a case across the handoffs
Choose an ordinary case and one exception. For each step, record the input, action, output and person receiving it next. Ask the receiving person what they need to accept the work. The answer often reveals assumptions that the sender never knew they were making.
Keep elapsed waiting time separate from active effort. A ten-minute task may wait two days for clarification. Speeding up the task does not necessarily resolve the delay.
Add the judgment behind the step
At each handoff ask: What decision is being made? What evidence supports it? Who recognizes an unusual case? Where does that person’s knowledge come from? What happens when they are unavailable?
Include workarounds and informal relationships with the participants’ knowledge. A map built only from formal roles can miss the planner everyone calls, the service specialist who recognizes a recurring fault or the colleague who translates between teams.
A worked example: an urgent quotation
In this illustrative manufacturer, sales captures a request, engineering confirms the specification, and planning checks capacity before sales releases a quote. The exception is an urgent order with an ambiguous component requirement.
Engineering knows the substitution may be acceptable, but planning needs to understand its production implications. A senior planner calls an experienced operator before committing. The system records the final date, while the reasoning and relationship behind it remain hard to find.
That creates several possible interventions: clearer input requirements, a documented substitution decision, a backup reviewer or a better way to capture confirmed context. AI could help with a suitable subtask, but the map does not establish that automation is the best first change.
Use the map to support people
Invite the people involved to review the account. Ask what would make contributing worthwhile: fewer repeat questions, recognition of experience, easier handovers or access to learning that helps with difficult cases.
Map dependencies to improve continuity, not to label someone a bottleneck without understanding their workload. Separate a confirmed capability from an inference about a person. Record uncertainty and provide a way to correct it.
What a useful map contains
Each row needs a step, decision, evidence source, responsible role, receiving role and exception route. Add an owner and review date so the map can stay useful. For a first exercise, a shared table is enough.
The wider opportunity is keeping this context connected as work changes. Gobekli calls that design the Human Intelligence Layer. TalentSync is intended to connect organizational workflows, experience and capabilities, while TalentPass supports the individual side of selected evidence and sharing.
An active Talent Tree would connect capabilities used in the work with relevant experience and gaps. It is more than a list of job titles. Integrations, ongoing conversations and review processes must be scoped against available capabilities before promising a working deployment.
Choose one change to examine
Use the map to pick a decision the team can improve and identify the evidence needed to test it. Preserve the baseline and revisit the downstream handoff after the change. A cleaner diagram is useful only if it helps people do or understand the work better.
Try this with one client situation.
Copy these prompts into your working notes. Keep observations, assumptions and open questions separate.
- Step and work sample:
- Decision and evidence used:
- Responsible role and receiving role:
- Informal help or expertise:
- Exception route and backup:
- Participant correction:
- Proposed change, owner and review date:
Bring one client challenge.
Tell us the work you know, the method you use and the decision your client needs to make. Let’s explore a useful first engagement.
Compare notes with GobekliExplore AI consultant partnershipsImplementation capabilities and support are agreed for each engagement. Review product availability.