卖什么
Open source self-improving layer for AI agents
商业模式拆解
买家：谁掌握预算？
未核实Engineering leads or founders at teams shipping LLM agents; signup is self-serve CLI/cloud, so individual engineers can start it.
触发点：什么让他们现在购买？
Agents pass tests but fail in production (wrong answers that look like successful runs, tool loops wasting tokens, regressions after prompt or model changes).
工作流：什么输入变成什么输出？
Trace production model and tool calls; LLM-judge detectors flag bad runs; judge outputs cluster into signals; a background coding agent investigates traces and GitHub code in a sandbox and recommends fixes; apply with your coding agent and verify with offline evals.
经济账：省了什么，怎么收费？
Open-source core (Apache 2.0) with paid cloud: usage-based on events/month, chat runs, RCA runs, with overage charges; enterprise license for ee directories.
渠道：怎么接触到客户？
Self-serve: npm CLI install, free cloud plan with no credit card, GitHub open source (796 stars), Discord community; Contact Sales for Enterprise.
融资历史与报道
| 融资日期 | 轮次 | 金额 | 投资方 | 来源 |
|---|---|---|---|---|
| — | YC 2025 夏季 | — | Y Combinator | ycombinator.com |
档案汇总自公开报道和公司官网。商业模式部分是 MaxBC 对这些材料的解读，不是已确认的事实；未知之处留空，不作猜测。 更新于 2026-10-12.