This week’s guest is Brooke Hopkins, Founder of Coval, a company building automated testing infrastructure for AI agents starting with voice and chat assistants.

We unpack her journey from studying in Abu Dhabi to leading simulation infrastructure at Waymo, where she helped ensure autonomous vehicles could operate safely at scale. Brooke shares how that experience shaped Coval’s approach to agent evaluation, why voice is the hardest modality to test, and what it means to simulate human-to-agent interactions in a way that actually reflects user intent.

Episode Chapters:

1:54 - Abu Dhabi for college
4:06 - Searching for engineering challenges
5:08 - Joining Waymo
9:00 - Scaling simulations
10:17 - Balancing cost, latency, and signal
14:40 - Starting with voice
18:16 - Building evals
23:44 - Persona surprises
26:50 - MCP
27:15 - AI operating system for conversational workflows
32:28 - Voice to voice
35:55 - Quick fire round

As always, feel free to contact us at partnerpathpodcast@gmail.com. We would love to hear ideas for content, guests, and overall feedback.

This episode is brought to you by Grata, the world's leading deal sourcing platform. Our AI-powered search, investment-grade data, and intuitive workflows give you the edge needed to find and win deals in your industry. Visit grata.com to schedule a demo today.

Fresh out of Y Combinator’s Summer batch, Overlap is an AI-driven app that uses LLMs to curate the best moments from podcast episodes. Imagine having a smart assistant who reads through every podcast transcript, finds the best parts or parts most relevant to your search, and strings them together to form a new curated stream of content - that is what Overlap does. Podcasts are an exponentially growing source of unique information. Make use of it! Check out Overlap 2.0 on the App Store today.