Kiriti Badam

AI/ML engineer at OpenAI working on Kodex (OpenAI’s coding agent). Spent a decade building AI and ML infrastructure at Google and Kumo before joining OpenAI. Co-teaches the top-rated AI product course on Maven with Aishwarya Naresh Reganti.

Work covered in this wiki

• Aishwarya and Kiriti on AI Products — Lenny’s Podcast, 2025. Agency-control trade-off; CCCD framework; evals; multi-agent systems; 2026 predictions.

Key ideas

• “Pain is the new moat” — companies winning at AI are those that went through the pain of understanding their specific workflows and failure modes; that accumulated knowledge is the real competitive advantage.
• Multi-agent systems are misunderstood — peer-to-peer gossip protocols between agents don’t work; supervisor-subagent patterns work; human-orchestrated multi-agent pipelines work.
• Coding agents are underrated in terms of global enterprise penetration — most value capture still ahead.
• Background/proactive agents as the key 2026 direction — agents that understand your workflow context and anticipate actions before being asked.

External

• LinkedIn: Kiriti Badam