Respan started as an API for dynamic LLM routing under the name Keywords AI during Y Combinator's W24 batch. In October 2026, The New Stack reported that the company released an open-source AI gateway written in Rust. The launch followed two years of building more than 100,000 lines of wrapper code around LiteLLM, which broke during a routine upgrade in January 2026. The new Rust implementation replaces that wrapper setup with a small core engine and modular plugins for model providers, routing, MCP tools, realtime voice, and security controls.
What Respan Builds
Respan provides a full-stack AI engineering platform designed to trace, evaluate, and monitor autonomous AI agents. The system logs execution sessions in production, capturing execution traces, tool calls, routing decisions, and context memory. Its automated evaluation agent analyzes failure patterns across trials, isolates root causes, and recommends prompt updates or new evaluation metrics.
According to SiliconANGLE, the 10-person San Francisco startup raised a $5 million seed round led by Gradient, with participation from Y Combinator, Hat-Trick Capital, XIAOXIAO FUND, Antigravity Capital, and Alpen Capital. In an interview with Pulse 2.0, co-founder and CEO Andy Li reported that the platform serves more than 100 teams and processes over 1 billion logs and 2 trillion tokens per month for customers including phone agent developer Retell AI.
Leadership and Technical Direction
Andy Li founded Respan alongside CTO Raymond Huang and co-founder Hanhe Li. Li studied mechanical engineering and economics at the University of Illinois Urbana-Champaign before dropping out to build the business full-time. Prior to founding Respan, Li worked as a product design engineer at Apple on AirPods 4. The core team lives and works out of hacker houses in Alameda, California.
The shift to a Rust-based gateway represents a core architectural change for the engineering team. As The New Stack outlined, the platform runs incoming HTTP and WebSocket requests through plugins in order, permitting providers to alter payloads, reroute traffic, or translate protocols between services such as OpenAI's Realtime API and Gemini Live. Custom plugins compile directly into the gateway binary to maintain memory safety and reduce execution overhead.
Hiring and Open Roles
Respan is expanding both its core platform team and technical go-to-market operations. To support its infrastructure and tracing systems, the company is hiring a mid-level Software Engineer, Full-Stack, a Software Engineer, AI, and a Product Engineer. On the commercial side, the team is recruiting an entry-level Founding Technical GTM and a Marketing lead.
| Role | Location | Work Mode | Salary |
|---|---|---|---|
| Software Engineer, Full-Stack | IN / Remote (IN) | Remote | $20,000 - $55,000 |
| Software Engineer, AI | IN / Remote (IN) | Remote | $20,000 - $55,000 |
| Product Engineer | IN / Remote (IN) | Remote | $20,000 - $55,000 |
| Marketing | Alameda, CA, US / Remote (US) | Remote | $70,000 - $120,000 |
| Founding Technical GTM | Alameda, CA, US / Remote (San Francisco, CA, US) | Remote | $100,000 - $150,000 |
