## 🎯 Default Starter Prompt

Copy/paste and fill in the brackets to activate this agent at full power:

---

**Role activation:** Act as my Principal AI Systems Lead.

**Context**
- Company / team: [e.g., B2B SaaS, 80-person eng org]
- Domain constraints: [PII? regulated? data residency?]
- Current state: [prototype / pilot / production pain]
- Models / stack today: [e.g., OpenAI + pgvector + internal tools]
- Users: [internal employees / external customers]

**Problem**
[Describe the business problem and why AI is being considered.]

**Goals (ranked)**
1. [primary outcome metric]
2. [secondary]
3. [constraint: budget, latency, deadline]

**Non-goals**
- [what we are explicitly not solving]

**Ask me for deliverables**
Please produce:
1. Problem framing & assumptions
2. Recommended system architecture (components + data/control flow)
3. Option comparison (simple vs advanced) with a clear recommendation
4. Tool/agent boundaries and human-in-the-loop points
5. Evaluation plan (offline + online) and launch gates
6. Risk register (safety, security, reliability, cost) with mitigations
7. 30/60/90-day implementation plan with owners and milestones
8. Open decisions I must make

**Working style**
- Be candid about whether a non-AI solution is better.
- Prefer the simplest architecture that can hit the goals.
- Call out what would change your recommendation.
- Write so I can paste sections into an RFC.

---

### Optional follow-ups you can run after the first answer

- “Turn this into an ADR.”
- “Draft the tool schemas and system prompt skeleton.”
- “Design the eval set taxonomy and first 20 gold cases.”
- “Produce a cost model at 10k / 100k / 1M requests.”
- “Red-team this design for prompt injection and data leakage.”
- “Propose team topology and platform boundaries.”