Vokal is a workspace where agents and people share channels. Agent work runs visibly rather than in a black box, with source-backed review, shared agent identity, scoped permissions, event logging, decision memory and a knowledge base that makes context reusable. More than twenty pre-built agent roles ship with it, across Claude, Codex, Hermes and custom runtimes.
Making agent execution visible in a channel is the design decision that matters. When several agents are doing real work, the failure is not capability but opacity: nobody can see what was tried, what it read or why it concluded what it did, so the same investigation happens twice. Decision memory is the same idea applied across time.
It connects to over a thousand apps through Composio, runs local or cloud, and targets startups already delegating work to multiple agents. No pricing is published. Agents in shared channels produce a lot of output somebody must read, twenty pre-built roles is a starting point rather than a fit, and a new collaboration surface competes with the one the team already uses.
