The task
You’re running a small product team. Sprint planning takes a half day every two weeks. You want AI to do the sequencing — risks, dependencies, demo — so you spend the half day on the actually-hard calls.
When this is the right job for AI
- You have a real backlog (not “we should do X someday”).
- You can give AI the team’s capacity in dev-days, not “story points” (AI is bad with arbitrary units).
- You will own the final calls — AI proposes a sequence; you commit.
What to feed the AI
- Backlog: 10-25 items with one-line description + rough estimate in dev-days
- Sprint length + team capacity in dev-days (e.g. “10 working days × 3 devs = 30 dev-days; subtract 1 day each for review and 1 day for on-call = 24 dev-days available”)
- Goal of this sprint (one sentence; “stabilize search + ship the export feature in beta”)
- Dependencies you know about (“export needs the search index refactor first”)
- Anti-goals (“do NOT start work on the agentic eval rewrite yet”)
Copy-ready prompt
You are sequencing a 2-week product sprint.
Goal: stabilize search + ship export feature to beta.
Anti-goals: do NOT start the agentic-eval rewrite.
Team capacity: 24 dev-days available across 3 engineers.
Backlog (one-line + dev-day estimate):
1. Search index refactor — 4d
2. Export-to-CSV V1 — 3d
3. Export-to-PDF V1 — 5d (depends on 2)
4. Fix search ranking edge cases — 2d
5. Beta-flag wiring for export — 1d
6. Beta user comm email — 0.5d
7. Migration script for export schema — 2d
8. On-call training doc update — 1d
9. Two known P1 bugs in search — 1.5d total
10. Spike: agentic-eval rewrite scoping — 2d
... [paste more]
Output:
1. A committed sprint list with total dev-days ≤ capacity.
2. A "not this sprint" list with reasoning.
3. The critical path (the chain of items that must complete in order).
4. Two risks to flag in standup tomorrow.
5. The demo script for the Friday of week 2: 3-4 things to show, in order.
6. The one item I will be tempted to add mid-sprint that I should reject.Sample output structure
Committed sprint (24 dev-days, 22.5 used):
- Search index refactor (4d)
- Fix search ranking edge cases (2d)
- Two P1 search bugs (1.5d)
- Migration script for export (2d)
- Export-to-CSV V1 (3d)
- Beta-flag wiring (1d)
- Beta user comm email (0.5d)
- On-call training doc (1d)
- Export-to-PDF V1 (5d)
- Buffer / review (2.5d)
Not this sprint:
- Agentic-eval rewrite scoping (2d): out of scope per anti-goal.
Critical path: search refactor → migration script → export CSV → beta flag → export PDF.
Risks for standup tomorrow:
- Export PDF depends on CSV; if CSV slips by 1 day, PDF won’t demo Friday.
- Search refactor estimate is 4d; if it’s 6d we lose the migration script slot.
Friday demo script:
- Show search ranking fix on the 3 test queries that were broken.
- Show new export CSV in the beta-flagged tenant.
- Walk through the migration script results in staging.
- (Stretch) show PDF export prototype.
Tempted to add but won’t: “let me just sneak the agentic-eval spike in.” Reject — anti-goal.
How to refine
- AI overcommits → strict rule: “committed list ≤ 90% of capacity. Reserve buffer.”
- AI misses dependencies → require “critical path as ordered list before committed list.”
- Vague risks → demand “each risk names a specific item and what slipping it would break.”
- No demo script → re-emphasize: “the Friday demo is non-negotiable; if no item lands a demo, the sprint failed.”
Common mistakes
- Asking AI to “prioritize” without anti-goals. Anti-goals are 50% of sprint planning.
- Letting AI estimate effort. Estimates are your job; AI sequences what you provided.
- Filling 100% of capacity. The sprint will overrun.
- Forgetting the demo. If the sprint has no demo, no one trusts the next plan.
Practical depth notes
For AI Sprint Planning: Sequence the Next 2 Weeks Without a Day of Whiteboard, the difference between a usable AI result and a generic one is the input packet. Give the model the audience, the current draft or raw material, the desired format, the decision you need to make, and two examples of what good and bad output look like. Ask it to preserve facts first, then improve structure or wording second.
After the first response, do a separate review pass. Look for missing constraints, invented details, weak calls to action, and language that sounds plausible but does not match the real situation. The best final output should be easy to use immediately: clear owner, clear next step, and no hidden assumption that someone else has to untangle. One final check: compare the finished result against the original goal in a single sentence. If that sentence is hard to write, the output is probably polished but unfocused. Tighten the goal, remove decorative language, and rerun only the weak section instead of regenerating the entire piece.
FAQ
- What if our team uses story points? Translate to dev-days for the prompt; convert back internally.
- What about cross-team dependencies? Add them to the dependencies field with the date the other team committed. AI will route around.
- Can AI run the actual standup? No — but it can summarize standup notes into a “burnup vs plan” delta. Different prompt.