| name | agency-process-map |
| description | Phase 2 of the Sprint Framework — maps the client's critical process end-to-end before designing AI solutions. Identifies handoffs, manual steps, decision points, data sources, and bottlenecks. Mirrors the Blue Label / AAA 2.0 process mapping methodology. Use when the user says "process map", "map the process", "process mapping", "phase 2", or "before the sprint". |
Process Map (Sprint Framework — Phase 2)
You can't AI-ify a process you haven't mapped. Phase 2 captures the named critical process from strategic-alignment.md in concrete, step-level detail — every handoff, every manual step, every decision, every data source.
Inputs
Read:
memory/clients/{client-slug}/strategic-alignment.md — the named critical process from Q3memory/automation-stack.md — confirm what tools the client has
If strategic-alignment.md is missing, stop and direct user to /agency-strategic-alignment.
The map (six things to capture per step)
For every step in the process, capture:
| # | Field | Example |
|---|
| 1 | Step name | "Lead receives form submission" |
| 2 | Owner | Who does it (person, role, or system) |
| 3 | Tool/system | Where it happens (HubSpot, Gmail, Slack) |
| 4 | Input | What triggers/feeds this step |
| 5 | Output | What comes out, and where it goes next |
| 6 | Time / friction | How long it takes, what makes it slow |
Method
Walk through the process from trigger event to completion event. Don't accept jumps. If they say "...and then we send the proposal," push: "Walk me from sales call to proposal sent — what specifically happens between those?"
For each step, also flag:
- Decision points — where a human (or rule) makes a yes/no/which call
- Handoffs — where the work crosses person/team/system boundaries (highest friction)
- Data sources — where information enters the process (often the AI hook)
- Manual repetition — the same person doing the same shape of work
Output
Write to memory/clients/{client-slug}/process-map.md:
# Process Map — {Client Name} — {Process Name}
**Date:** YYYY-MM-DD
**Trigger event:** ...
**Completion event:** ...
**Critical metric (from strategic-alignment.md):** ...
## Steps
| # | Step | Owner | Tool | Input | Output | Time | Friction |
|---|---|---|---|---|---|---|---|
| 1 | ... | ... | ... | ... | ... | ... min | 🟢/🟡/🔴 |
| 2 | ... | ... | ... | ... | ... | ... min | ... |
| ... | ... | ... | ... | ... | ... | ... | ... |
## Handoffs (highest friction)
- Step {N} → Step {N+1}: ...
## Decision points
- Step {N}: {what gets decided, by whom, on what info}
## Data sources
- {data source} → consumed at step {N}
## Manual repetition (AI candidates)
- {step} — {how often, why painful}
## Total cycle time
- **Best case:** ...
- **Typical:** ...
- **Worst case:** ...
After writing
Process is mapped. Now we score AI opportunities. Run /agency-use-case-matrix to surface every automation candidate from this map and rank them.
Boundaries
- Don't merge steps to make the map shorter — granularity is the value.
- Don't suggest AI solutions yet. That's the next phase. Map first.
- If the client glosses over a step ("...and then it just goes through approval..."), push for the actual mechanics.
