AI tooling watchlist — revisit triggers
Context
Tools that were "pass for now" but worth re-evaluating if a specific condition appears. Prevents both premature adoption and forgetting a genuinely useful option. Companion to external-ai-tool-evaluations + ai-tool-evaluation-rubric.
Detail
| Tool | Re-evaluate when | What would change the call |
|---|---|---|
| Sakana Fugu | hand-orchestration (OpenClaw + AIOS) becomes a proven bottleneck on hard multi-step tasks, AND Fugu matures past day-one | Offers learned multi-model orchestration we don't have today; it is OpenAI-compatible so it can trial behind the gateway. Watch metered cost vs the flat-rate stack. |
| Headroom | the fleet starts hitting Claude Max usage caps / throttling | Compression buys context headroom under the cap — value is throttle relief, not $ saved. |
| browser-use | a recurring no-API SaaS portal must be operated unattended at scale | Self-host the OSS lib behind the gateway; never the CAPTCHA/stealth cloud. |
Actions
- Review a row only when its trigger condition is actually observed (event-driven, not on a calendar).
- On adoption or final rejection, move the row into external-ai-tool-evaluations.
Provenance
Seeded 2026-06-22 from four back-to-back tool evaluations.
🔗 Relationships
graph LR
ai_tooling_watchlist["ai-tooling-watchlist"]:::self
ai_tooling_watchlist --> external_ai_tool_evaluations["external-ai-tool-evaluations"]
ai_tooling_watchlist --> ai_tool_evaluation_rubric["ai-tool-evaluation-rubric"]
classDef self fill:#715EE3,color:#fff,stroke:#291F50;