Most browser agents take a screenshot and guess where to click. rtrvr.ai doesn't take screenshots at all. In this episode of Agent Hour, Gabi Weinberg sits down with Arjun Chintapalli and Bhavani Kalisetty, the ex-Google engineers who bootstrapped rtrvr.ai from a Chrome extension into a cloud browser platform and an in-product onboarding agent. Their argument is technical: language models are language models, extraordinary at semantic understanding, mediocre at finding a pixel coordinate. So rather than train a vision model, rtrvr.ai compresses a page's DOM, which can run into millions of tokens, into a representation that preserves every possible action, with element IDs mapped to real on-page text so the agent has far less room to invent things. It also runs on any off-the-shelf model, including text-only ones like DeepSeek; cheap enough, they argue, that a single ad impression could cover an entire agent task.

🎧 Brought to you by Anchor Browser — cloud-native browser infrastructure for AI agents.

Evan Barnard runs AgentGrail, an AI-powered sports and trading card intelligence platform that scans live eBay listings around the clock and calls BUY, PASS, or REVIEW using computer vision and price/grading reasoning trained on thousands of graded slabs. On this episode of Agent Hour, Evan joins to talk about why a niche, high-emotion hobby market turned out to be a great testbed for autonomous agents, what breaks trust when a bot's buy call is wrong and real money's on the line, and what generalizes from trading cards to agentic commerce everywhere else.

🎧 Brought to you by Anchor Browser — cloud-native browser infrastructure for AI agents.

When an agent's session ends, its context usually goes with it — and that problem multiplies fast once you've got multiple agents running at once, each with different access to different data. Vas, founder and CEO of Cognee, joins Agent Hour to explain why simple vector search isn't enough for real agent memory, and how knowledge graphs let agents actually reason over relationships instead of just retrieving similar chunks. He walks through Cognee's ECL pipeline (Extract, Cognify, Load) for turning messy multi-source data into something agents can navigate, plus how granular, per-agent access control keeps each agent's "world model" clean as teams scale.

🎧 Brought to you by Anchor Browser — cloud-native browser infrastructure for AI agents.

What happens when a company stops relying on a shared Notion doc to remember how it works — and lets AI agents do that instead? This week, Gabi is joined by Jean-Denis Greze, co-founder and "Mayor" (not CEO) of Town, an AI assistant startup that just raised a $55M Series A led by Andreessen Horowitz. Before Town, Jean-Denis spent nearly seven years as CTO of Plaid, scaling the engineering org from 20 people to 350 through a $13.4B valuation. He breaks down his thesis of the "self-organizing company," how Town's agents read across email, Slack, and docs without eroding trust between coworkers, and why — despite building an AI assistant for a living — he still won't let his own agent send an email on his behalf.

Guest: Jean-Denis Greze, Co-founder & CEO & "Mayor," Town
Links: town.com

🎧 Brought to you by Anchor Browser — cloud-native browser infrastructure for AI agents.

Most companies rolling out AI agents have no idea how much "shadow AI" is already running inside their org. Peter Holcomb, founder & CEO of Optimo AI and fractional CISO (formerly at Datavolo, acquired by Snowflake, and eMed), joins Agent Hour to break down what agent identity and access actually looks like in practice, the compliance gaps that kill enterprise deals, and who's accountable when an agent gets it wrong. Drawing on 100+ engagements helping AI-native companies get through SOC 2, ISO 27001/42001, GDPR, and HIPAA, Peter shares the governance playbook most teams don't build until it's too late.

🎧 Brought to you by Anchor Browser — cloud-native browser infrastructure for AI agents.

Antonio Mallia is the founder of Seltz.ai, building search infrastructure from the ground up for a world where the reader is increasingly an agent, not a human. Before Seltz, he spent time at Bloomberg, Amazon's AGI team, and Pinecone, giving him a front-row seat to how retrieval breaks down at scale. In this episode, we dig into why he chose to own the full stack instead of wrapping an existing search index, how trust and authority get scored when there's no human eyeballing a source, and what happens to publishers and the open web as bots increasingly do the reading. We also get his rapid-fire take on what's overrated and underrated in AI search infrastructure today.