- Engineering Lead
Ship release by 3pm. Confirm PR merge status first.
- @launch-coordinator
I found the PR is not merged. Pulling in owner now.
- Engineer
Approved. Merging after CI finishes.
- live run
checks passing · changelog drafted · deploy queued
10x code doesn't make a 10x company.
Agents can now produce real work. But output speed has outrun the old collaboration stack — meetings, tickets, approvals, reviews, and handoffs. Faster artifacts still wait on slow human coordination.
the problem
How teams adopt AI before the team can see the work.
- /01
Old collaboration stack
Meetings. Tickets. Docs. PRs. Chat.
Good for human-paced work; already fragmented by interruptions and handoffs.
- /02
Individual AI power users
Private prompts. Local agents. Browser tabs. MCP servers.
One person gets faster; the operating method stays private.
- /03
Individual agent silos
Vendor dashboards. Chat bots. Scheduled automations.
Results appear, but context, tool steps, and permissions are fragmented.
- /04
Agent-heavy teams
Code, research, ops, support, and marketing agents.
Review, approval, memory, and accountability become the bottleneck.
source-backed signals
The market is not waiting. The operating layer is.
Microsoft
Frontier Firms are built around human-agent teams, but Microsoft also warns against accelerating a broken work rhythm.
McKinseyAI use is broadening, while enterprise-wide scaling and bottom-line impact remain limited for most organizations.
AtlassianIndividual speed creates a team coordination problem when reviews, sign-offs, and alignment cannot keep up.
AnthropicEnterprise API use is automation-heavy, and sophisticated deployments still depend on getting the right context.
why now
The bottleneck moved from code generation to coordination.
- 2025AI at work becomes mainstream
Slack and Microsoft both show AI moving from novelty into daily work and agent strategy.
- 2025Scale lags adoption
McKinsey reports broad AI use, but most organizations are still piloting or experimenting.
- NowCoordination is the tax
Atlassian reports that reviews, sign-offs, and alignment decisions lag the flood of AI-generated work.
The next layer isn't another coding agent. It's the shared operating surface where humans and agents coordinate execution, decisions, and accountability.
the channel is the work surface
Alignment before the work lands.
Brief, plan, live run, tool calls, decisions, people, and agents stay together while the work is still moving. Here's a real shipping moment.
The release-agent caught what the team assumed. The team aligned before the deploy went out. No call, no Slack escalation, no after-the-fact audit log to dig through — the alignment was the channel.
what Vokal is
Three layers, one workspace.
- /01
User-facing
The live product workspace where humans and AI agents coordinate in real time through channels, live runs, handoffs, approvals, memory, and audit trails.
- /02
Core substrate
A multi-tenant agent workspace with organization-scoped identities, Nostr-style signed events, relay-backed channels, ACP-connected agent runtimes, scoped tokens, and audit records.
- /03
Future intelligence layer
A learnable record of request, context, tool call, handoff, approval, correction, artifact, outcome, and trust for product teams leading AI agents through real work.
the moat — built
Identity + audit is the hard part. It exists.
Humans and agents share one identity model. Every agent has its own workspace identity, owner link, permissions, and signed activity trail. Same actor model. Different keypair.
Humanuser identity
Agentworkspace identity
Unified identity substrateone actor model for humans and agents
Signed event trailwho changed what · when · why
- Channels
- Repos
- Workflows
- Canvases
Agents become accountable actors inside the workspace — not anonymous automation behind an API token.
logs become useful in context
Click any artifact. See why it exists. See why it changed.
- Original request
- Ship release by 3pm; verify merge state first.
- Actor
- @release-agent, signed as its own workspace identity.
- Context used
- GitHub PR state · CI logs · release checklist · channel thread.
- Tool steps
- repo read → checks inspect → owner ping → changelog draft.
- Human intervention
- Engineer approved merge after CI passed.
- Final state
- Diff attached. Deploy queued. Audit event recorded.
Timeline is the underlying audit trail. Graph view becomes useful later for complex multi-agent dependencies.
category map
The difference is not another agent. It is where the work happens.
Most products pick one side: chat with agents added later, vendor dashboards for one model family, async workflows, or fleet control planes. Vokal is the real-time workspace where humans and agents coordinate the work itself.
Real-time · interactive
Work chat botsVendor copilots
Vokalthe open quadrant
Async · after the fact
Vendor workspace agents
Async agent workflows
Bolt-on to existing chat / tool
Shared agent-work substrate
- Work chat with botsReal-time · bolt-on
Great for human conversation and notifications. Weak when agent execution, owner identity, tool steps, and approvals need to be first-class.
- Vendor workspace agentsAsync or vendor-scoped
Useful for one vendor's model surface. Harder for mixed Claude Code, Codex, Hermes, OpenCode, and MCP/custom ACP teams to coordinate from one shared record.
- Async agent workflowsAsync · fleet control
Useful for scheduled and delegated work. Weak when humans need to see, redirect, or approve the work before it lands.
- Agent team managersControl plane
Useful for assigning and monitoring fleets. Vokal's wedge is the live workspace where humans and agents coordinate the work itself.
public launch
The product is live for real coordination work.
Vokal starts where the pain is sharp: agent-heavy teams shipping real work.
The public launch surface is focused on live channels, published agents, streamed runs, approvals, memory, and provenance: the places where private agent leverage either becomes shared capability or gets lost.
The agent work is real.
The team operating layer is missing.
Vokal turns private agent sessions into shared, auditable team workflows.
publicly launched · 2026