
Build Mode Building a lean team before raising big with David Park, Narada
11 snips
Mar 5, 2026 David Park, co-founder and CEO of Narada, a Berkeley-spun enterprise AI agent founder who previously built and exited Coverity. He discusses using large action models to automate complex workflows. He talks about spending a year on nearly 1,000 customer calls, achieving 99.99% production reliability, why raising too early can derail teams, and how to build a lean, disciplined go-to-market approach.
AI Snips
Chapters
Transcript
Episode notes
Enterprise Needs 99.99% Reliability Not Lab Demos
- Enterprise automation requires near-deterministic reliability, not best-effort probabilistic outputs from LLMs.
- Narada focuses on engineering for 99.99% consistency by solving many last-mile production issues rather than relying on a single smarter base model.
Drop Data Post Task And Offer On Prem Deployments
- Security and data policy shape enterprise deployment: Narada drops data after task completion and offers on-premises deployment.
- This combination reassures regulated industries like finance and insurance that nothing leaves their perimeter.
Automate The UI To Avoid Costly Customer Rewiring
- Minimize customer-side changes by building agents that operate the UI like humans instead of requiring backend integrations.
- Narada asks customers to show the manual process and automates it by reading screens, clicking, typing, and calling APIs when available.
