Generating Work Solutions
Given a problem statement, produce 3-5 distinct solution options, each with approach, effort, risk, and recommendation — not just one "correct" answer.
Problem: "Support tickets are piling up; team can't keep SLA."
Option A — Triage automation (Low effort, Fast impact)
Option B — Hire temp support, 2 FTE (Medium effort, Medium-term fix)
Option C — Self-serve knowledge base (High effort, Long-term fix)
Option D — Renegotiate SLA with key clients (Low effort, Political risk)
Recommendation: A now + C in parallel, revisit B if backlog doesn't drop in 2 weeks.
Progress:
- Step 1: Clarify the real problem (symptom vs root cause)
- Step 2: Identify constraints (budget, time, people, authority)
- Step 3: Generate options across different axes (fast/cheap vs slow/robust, manual vs automated, internal vs outsourced)
- Step 4: Evaluate each option (effort, cost, risk, time-to-impact, reversibility)
- Step 5: Recommend a path (often a combination, with sequencing)
- Step 6: Note what would invalidate the recommendation (assumptions to watch)
Step 1 detail — clarify the problem: Ask (internally or explicitly): Is this a one-time problem or recurring? Is the stated problem the actual bottleneck, or a symptom? What does "solved" look like concretely?
Step 3 detail — force diversity: Don't generate 5 variations of the same idea. Deliberately pull from different categories:
- Quick patch vs structural fix
- Do less (cut scope) vs do more (add resources)
- Change the process vs change the people/tools
- Buy vs build vs eliminate the need entirely
Step 4 detail — evaluate consistently: Use the same dimensions for every option so they're comparable: effort/cost, time to see impact, risk if it fails, reversibility (easy to undo?), dependencies (who else needs to approve/act).
Example 1: Input: "Our weekly team meeting is unproductive — people check phones, nothing gets decided." Output:
- Option A: Cut meeting to 15 min, standup format, decisions only (Low effort, fast)
- Option B: Pre-read doc required, meeting becomes Q&A + decisions (Medium effort)
- Option C: Replace with async updates in Slack + biweekly sync (Higher effort, cultural shift)
- Option D: Rotate facilitator + mandatory agenda with owner per item (Low effort) Recommendation: Combine A + D immediately (no cost, fast test); escalate to C if attendance/engagement doesn't improve in 3 weeks.
Example 2: Input: "We keep missing deadlines because requirements change mid-sprint." Output:
- Option A: Lock requirements at sprint start, changes go to next sprint (Process fix, may cause friction with stakeholders)
- Option B: Shorten sprints to 1 week to reduce drift window (Structural, needs tooling adjustment)
- Option C: Add a "change buffer" — reserve 20% capacity for mid-sprint changes (Pragmatic, immediate)
- Option D: Root-cause stakeholder behavior — why do requirements change so often? (Root-cause fix, slower) Recommendation: C as immediate relief, D in parallel as the real fix; A only if C fails.
- Always give more than one option — a single answer invites binary accept/reject instead of informed choice.
- Make trade-offs explicit (effort vs impact vs risk) instead of just listing ideas.
- Separate symptom relief from root-cause fixes and recommend both where relevant.
- End with a clear recommendation, not just a neutral list — decision-makers want a default, not homework.
- Flag assumptions: state what would change the recommendation if untrue.
- Giving only variations of one idea instead of genuinely different approaches.
- Recommending the most resource-intensive "best" solution without a low-cost immediate option.
- Ignoring stated constraints (budget, authority, timeline) when generating options.
- Skipping the recommendation step and leaving the reader to choose from an undifferentiated list.
- Treating every problem as needing a new process — sometimes the fix is removing a step, not adding one.