Stripe's Internal AI Assistant Kai: 83% Weekly Adoption and Engineering Lessons
xiaohu · x · 2026-08-05
Stripe built an internal AI assistant named Kai using LangChain's open-source Deep Agents, achieving 83% weekly employee adoption. The teardown shares valuable engineering insights and pitfalls in building enterprise agents.
- Architecture: The first working version was built by a single engineer in one week. It heavily leveraged off-the-shelf components for files, sandboxing, and context compression, while custom-building the security boundaries.
- Capabilities: Pre-loaded with internal company knowledge and over 1,000 skills/tools, Kai allows non-technical staff to generate reports and dashboards via natural language.
- Counter-intuitive Finding: Testing revealed that equipping the agent with more than 150 skills degrades performance, as frontier models struggle to select the correct skill without errors.
Related event: Stripe Built Internal AI Agent Kai in One Week with Single Engineer(4 posts)→
More from coding & agent
- Unicity Launches Multi-Tenant Agent OS with 1000x Density — JoshuaJBouw · 2026-08-05
- mattpocock/skills Launches New Docs for AI Engineering Workflows — mattpocockuk · 2026-08-05
- mattpocock/skills v1.2 Released: New Slash Commands for AI Coding — mattpocockuk · 2026-08-05
- Dev Exhausts Codex Credits After 100-Hour Reverse Engineering Spree — yacineMTB · 2026-08-05
- OpenAI Agents Repo Skill Offers Risk-Tiered Code Review to Improve First-Pass Quality — gabrielchua · 2026-08-05
- Training Coding Agents with RL: OpenCode Harness in HF Sandboxes — SergioPaniego · 2026-08-05