| name | agency-use-case-matrix |
| description | Phase 2.5 of the Sprint Framework — surfaces and scores every AI automation candidate from a process map, then selects the best pilot. Mirrors the Blue Label / AAA 2.0 use-case matrix methodology. Use when the user says "use case matrix", "AI opportunities", "score automations", "pilot selection", "phase 2.5", or after process mapping. |
Use Case Matrix (Sprint Framework — Phase 2.5)
The bridge between a mapped process and a sprint plan. Looks at every step in the process map, surfaces every AI / automation opportunity, scores them, and picks the pilot.
Inputs
Read:
memory/clients/{client-slug}/strategic-alignment.md — the metric we're movingmemory/clients/{client-slug}/process-map.md — the steps, handoffs, manual repetition
If process-map.md is missing, stop and direct user to /agency-process-map.
Surface candidates
For each row in the process map, ask three questions:
- Is there manual repetition or judgment that an LLM/agent could perform? (yes/no)
- Is the input data structured enough for AI to act on? (yes/partly/no)
- What's the failure mode if AI gets it wrong? (low/medium/high)
Anything answered "yes" to Q1 becomes a candidate. List them all — don't filter yet.
Score each candidate (4 dimensions)
| Dimension | What it measures | Score 1-5 |
|---|
| Impact on the metric | How much does this move the Q2 metric? | 1=barely / 5=most of the gap |
| Implementation effort | Build complexity, integration depth | 1=hard / 5=easy |
| Time to value | How fast can we prove it works? | 1=slow / 5=fast |
| Risk of failure | What's the blast radius if it breaks? | 1=high / 5=low |
Pilot score = sum of the four (max 20).
The matrix
LOW EFFORT HIGH EFFORT
HIGH IMPACT ┌─────────────────────────┐ ┌─────────────────────────┐
│ PILOT CANDIDATES │ │ ROADMAP (Now-Next-Later)│
│ → Sprint plan │ │ → Phase 2 retainer │
└─────────────────────────┘ └─────────────────────────┘
LOW IMPACT ┌─────────────────────────┐ ┌─────────────────────────┐
│ QUICK WINS / DEPRIORITIZE│ │ IGNORE │
│ → Backlog │ │ → Refer-out / kill │
└─────────────────────────┘ └─────────────────────────┘
Pilot selection
The pilot is the highest-scoring candidate that the client agrees on. Two rules:
- It must move the metric from
strategic-alignment.md. No exceptions.
- It must finish in the sprint window (typically 2 weeks). If it doesn't, downscope.
Ask the client: "Of these top 3, which one — if it worked — would convince you to do the next phase?"
That's the pilot.
Output
Write to memory/clients/{client-slug}/use-case-matrix.md:
# Use Case Matrix — {Client Name}
**Date:** YYYY-MM-DD
**Anchor metric (from strategic-alignment.md):** ...
## All candidates
| # | Step (from process-map) | Candidate | Impact | Effort | Time-to-value | Risk | Total |
|---|---|---|---|---|---|---|---|
| 1 | ... | ... | /5 | /5 | /5 | /5 | /20 |
| 2 | ... | ... | /5 | /5 | /5 | /5 | /20 |
| 3 | ... | ... | /5 | /5 | /5 | /5 | /20 |
| ... | ... | ... | ... | ... | ... | ... | ... |
## Top 3 (highest scores)
1. **{candidate}** — {one-line why}
2. ...
3. ...
## Selected pilot
**{Pilot name}**
- Why: ...
- Expected metric impact: ...
- Sprint window: 2 weeks
- Confidence: 🟢 High / 🟡 Medium / 🔴 Low
## Roadmap (high-impact / high-effort — not pilot)
1. ...
2. ...
## Backlog (low-impact / low-effort — quick wins later)
1. ...
After writing
Pilot selected. Now scope the sprint. Run /agency-sprint-planner to turn this pilot into a 2-week plan with daily milestones, success criteria, and tooling decisions.
Boundaries
- Don't pick the pilot the implementer wants to build. Pick the one the client will defend.
- Don't pad the matrix with weak candidates to look thorough. List only real ones.
- High failure-mode risk = downscope or pick something else. Don't pilot something that could damage the client business.
