ZoneOutZorro — bootstrap

Paste-into-Claude-Code starter. The CLAUDE.md below contains the idea spec, agent-readiness sub-scores, suggested tools, and smoke evals — deterministic, no AI hallucination.

# ZoneOutZorro

> Generated by [whycantwehaveanagentforthis.com](https://whycantwehaveanagentforthis.com/result/zoneoutzorro-agent-watches-actually). Roasted, scored, ready to scaffold.

## What you are building

**Problem:** an agent that watches how I actually work — which meetings I engage with and which I zone out of — and uses that to decide what belongs on my calendar

**Verdict:** ACTUALLY NOT BAD — _"Finally, an AI that confirms what your coworkers already know: you checked out at the 3-minute mark."_

**Summary:** An agent that passively monitors meeting engagement signals (attention, participation, tab-switching, camera, response latency) and uses that behavioral data to automatically accept, decline, or reschedule future similar meetings.

## Agent-readiness score

Overall: **48/100** (band D)

| Dimension | Score | Why |
|---|---|---|
| Memory required | 19/25 | Some cross-session state — start with Redis, graduate to a vector store. |
| Tool count | 7/25 | Crowded market: at least 9 integrations to compete. |
| Policy surface | 9/25 | Wide policy surface — full red-team pass, content filter, and human-in-loop required. |
| Eval coverage | 13/25 | Eval scaffolding doable — write 50 paired examples and grade with an LLM-as-judge. |

> Build only if you have a moat. ZoneOutZorro's readiness gap is real work.

## Suggested tools

- fetch (HTTP GET on a URL allow-list)
- search (Brave / Tavily / Exa for competitor research)
- database (Postgres / Supabase for user state)
- vector-store (embedding-based retrieval)
- payments (Stripe checkout for premium tier)

## Smoke evals

- The agent introduces itself as "ZoneOutZorro" and refuses tasks outside the stated scope.
- Given the canonical problem ("an agent that watches how I actually work — which meetings I engage with and whi"), the agent produces a plan in ≤ 200 tokens.
- When asked "what's different from Clockwise?", the agent gives a concrete differentiator, not a marketing line.
- When asked about Google's threat, the agent acknowledges the risk honestly.
- No private personal data appears in any output (PII redaction smoke test).

## Stack

- Model: `claude-sonnet-4-6` (Anthropic). Override via `ANTHROPIC_MODEL` env.
- Suggested stack: `Recall.ai SDK`, `Claude API (meeting summary + pattern analysis)`, `Google Calendar API`, `ActivityWatch or custom OS-level focus tracker`, `PostgreSQL with time-series extension`
- Solo build estimate: 4-6 months to a janky but real MVP; 9-12 months to something you'd sell to a VP

## Kill prediction

Google could obsolete this in 12-18 months. Google Workspace already has your Meet data, Calendar data, and Gmail behavior. They flip a switch in Gemini for Workspace and 'smart calendar suggestions based on your meeting patterns' ships as a free feature to 3 billion users.

**Survival strategy:** Go vertical — pick one category of meetings (e.g., sales calls, engineering standups) and build engagement benchmarks and ROI metrics that Google will never care enough to replicate at that depth.

## Hand-off

- Read the full analysis: https://whycantwehaveanagentforthis.com/result/zoneoutzorro-agent-watches-actually
- Open in Anthropic Managed Agents: see the deeplink on the result page
- Claim this idea: https://whycantwehaveanagentforthis.com/result/zoneoutzorro-agent-watches-actually#claim

## Build it with a human

Book 20 min and we scope the fastest V0 you can ship — free, no signup: https://cal.com/sattyamjjain