| name | agency-sprint-planner |
| description | Phase 3 of the Sprint Framework — turns a selected pilot into a 2-week sprint plan with daily milestones, success criteria, and tooling decisions. Mirrors the Blue Label / AAA 2.0 sprint planning methodology. Use when the user says "sprint plan", "plan the sprint", "phase 3", "design the pilot", or after picking a pilot from the use case matrix. |
Sprint Planner (Sprint Framework — Phase 3)
Turns the selected pilot from use-case-matrix.md into a concrete 2-week sprint plan. The output is the deliverable artifact you give the client at end of Phase 3 — and the document the implementer works against day-to-day.
Inputs
Read:
memory/clients/{client-slug}/strategic-alignment.md — the metricmemory/clients/{client-slug}/process-map.md — the workflow contextmemory/clients/{client-slug}/use-case-matrix.md — the selected pilotmemory/automation-stack.md — the agency's tools (what we'd build with)
If any are missing, route the user to the right phase.
What a sprint plan contains
- Sprint goal — one sentence. What does success look like at end of Day 14?
- Success criteria — measurable, ranked. The metric must be one of them.
- Daily milestones — what's done each day (or 2-day block)
- Tooling decisions — what we're building it on (Claude, Make.com, Custom GPT, n8n, etc.)
- Data requirements — what the client must provide, by when
- Risk register — top 3 risks + mitigation
- Decision points — moments the client must approve before we proceed
- Definition of done — when do we declare "pilot worked"?
Sprint shape (default)
A 2-week (10 working day) sprint. Adjust if pilot scope demands.
WEEK 1 — BUILD & VALIDATE
Day 1 Kickoff + data confirmation
Day 2 Architecture confirmed, scaffolding built
Day 3 Core agent / automation logic working in isolation
Day 4 First integration with a real data source
Day 5 Internal demo + Champion review
WEEK 2 — INTEGRATE & MEASURE
Day 6 Handoff with adjacent process steps
Day 7 Edge cases + failure modes handled
Day 8 Pilot runs against real client data (shadow mode)
Day 9 Side-by-side comparison vs. current process
Day 10 Pilot demo + Now-Next-Later roadmap (Phase 5 input)
Output
Write to memory/clients/{client-slug}/sprint-plan.md:
# Sprint Plan — {Client Name} — {Pilot Name}
**Sprint window:** YYYY-MM-DD → YYYY-MM-DD
**Implementer:** ...
**Champion:** ...
## Sprint goal
{one sentence}
## Success criteria (ranked)
1. **{Metric from strategic-alignment.md}** moves from X → Y
2. ...
3. ...
## Daily milestones
| Day | Deliverable | Owner | Decision needed? |
|---|---|---|---|
| 1 | ... | ... | ... |
| 2 | ... | ... | ... |
| ... | ... | ... | ... |
| 10 | Pilot demo + Now-Next-Later doc | ... | Go / no-go for retainer |
## Tooling
| Layer | Tool | Why |
|---|---|---|
| LLM | Claude Sonnet/Opus / GPT-4o / ... | ... |
| Glue | Make.com / n8n / Zapier / Claude Code | ... |
| Storage | Notion / Airtable / Postgres / ... | ... |
| Frontend | Custom GPT / web app / Slack bot / ... | ... |
## Data the client must provide
| Data | By when | Format |
|---|---|---|
| ... | Day 1 | ... |
## Risks
| Risk | Likelihood | Mitigation |
|---|---|---|
| ... | High/Med/Low | ... |
## Definition of done
✅ {Metric} measurably improved on real client data
✅ Champion + 1 user can demo it without the implementer
✅ Now-Next-Later roadmap delivered (Phase 5)
After writing
Sprint is planned. Send to the Champion for approval. When you're at Day 10 and the pilot is working, run /agency-now-next-later to map what's beyond the pilot.
Want this exported as a client-facing PDF? I can render the plan in their voice via /agency-sprint-proposal.
Boundaries
- Never plan a sprint longer than 14 days without explicit reason.
- Never let "tooling decisions" get pinned by the implementer's preferences — pin them by what actually serves the client's stack.
- If the success criteria don't include the metric from
strategic-alignment.md, you're planning the wrong sprint. Stop and re-anchor.
