/ Products

What we ship.

Current

Webcmd_

Self-learning browser infrastructure for AI agents

Webcmd learns how websites work as agents use them, then turns that knowledge into deterministic commands. Agents can control a live browser, reuse site memory, and graduate repeated workflows into reliable CLI adapters—cutting navigation overhead, token usage, and variance.

Live browser controlSite memoryCLI adaptersStructured outputOpen source

Open product →Source ↗

webcmd · deterministic weblive output

$webcmd hackernews top --limit 3

>Strategy: PUBLIC · browser: none

1Show HN: A local-first agent runtime

2Building reliable browser agents

3SQLite on the edge

✓3 rows · structured output · 412ms

>No page inspection or rediscovery needed

exit 0ready

▸Past projects2

Authsome_

Past

Credential broker and vault for AI agents

An open-source credential broker that sits between your agents and the services they call. Log in once via OAuth2 or API keys. Authsome stores credentials and injects them via an HTTP proxy, so you can rotate keys and see what every agent is doing from one place.

Open product →Source ↗

Ruzo_

Past

AI-first automation for go-to-market teams

Ruzo ran multi-step work from a single prompt—lead research, enrichment, and follow-through across the tools GTM teams already use. It was our first product, before the lab focused on agent infrastructure.

Open product →