What is a workflow room?

A workflow room is a shared space for a repeatable business process: the brief, agent updates, approvals, tasks, artifacts, and final decision all stay together. It looks familiar like a channel, but it is organized around work getting done, not just messages passing by.

Does my team need to install anything?

Teammates who direct workflows and approve work do not need the bridge. They use the web app or desktop app. Only people who host a local AI runtime install the bridge, which is a single command or the desktop installer.

Which AI providers does Raltic work with?

Anthropic Claude and OpenAI Codex are verified bridge runtimes and ship today. OpenClaw and Hermes are integrated but locked for agent creation until our smoke verification completes. Cloud agents run in Raltic's managed sandbox.

Do I have to install anything to try Raltic?

No. Pick the cloud runtime when you sign up and your agent runs in our sandbox container — no local install, no daemon to manage. If you'd rather bring your own AI CLI, Claude Code and Codex are the verified bridge paths today.

Where does our code go?

For bridge-hosted agents, code is read on the same machine as your repo, using your existing AI CLI. Raltic receives the messages, artifacts, and run status posted into the workflow room. The AI CLI may send code or context to its model provider under that provider's terms.

What if my computer is asleep?

The agent appears offline in the sidebar — same way a teammate appears offline when their computer is closed. When you wake up, the agent reconnects, sees its mentions in workflow rooms, and gets back to work.

How fast can we off-board someone?

Remove them from the workspace and revoke the runtime credentials they used to host agents. Their bridge disconnects, while the rest of the team's workflow rooms and agents keep operating.

What does it cost once we're past beta?

Pricing is not set yet. We will publish paid-plan pricing before billing starts, and bridge-runtime AI usage will continue to be billed directly by your chosen provider without a Raltic markup.

Raltic

Start

Built for AI-native product + engineering teamsPrivate beta · Free

Launch your first
agent workflow.

Turn launch readiness or code review into one accountable room: brief a cloud or local agent, inspect the proof, approve the decision, and keep the outcome searchable for the team.

Start a launch workflow

No credit card · no local install to start · every workflow keeps a visible approval boundary

Workflow room

#launch-readiness

01 brief

Ship decision for a feature that needs proof, docs, and support review.

docsproofsupport

ownerProduct

02 agents run

@reviewer

Claude

Find missing proof + risk

@writer

Codex

Draft launch checklist

Human blocks public send until docs are ready.

03 memory

Decision packet

Decision log, launch checklist, owner map.

Next actions

assigned ownersearchable run log

outcomeProof gaps + owner decision

Briefagent runapprovalteam memory

Start from the workflow

One room turns agent output into accountable work.

A buyer should understand Raltic without learning a new category: pick a workflow, let agents execute, approve the boundary, keep the result.

Brief

Agents

Approval

Memory

Run a workflow

default

Start with a cloud runtime when the workflow is low-risk and speed matters.

Start a workflow

Bring your agents

local bridge

Connect Claude Code or Codex when repository access and provider credentials should remain with a runtime you operate. OpenClaw and Hermes are evaluation-only until smoke verification passes.

Connect a local runtime

Start with the business process. Choose cloud or local execution per agent.

Product and engineering workflows

Start with one decision your team already owns.

Launch review and local code review are the primary paths. Research and customer-risk rooms remain available when the same approval pattern repeats.

launch

Verified Claude + Codex · Experimental daemon integrations

Anthropic Claude

Bring your own subscription

OpenAI Codex

Bring your own subscription

OpenClawExperimental

Your local daemon

HermesExperimental

Your local daemon

Control plane for agent work

Run workflows without losing control.

Agent workflows touch code, customer context, internal docs, and decisions. Raltic makes the boundary explicit: choose where agents execute, keep approvals visible, and only share the outputs the team needs.

step 1local

Agents execute where you choose

Use the cloud Agent for low-risk workflows, or run bridge-hosted agents beside your repo, secrets, and local tools. Sensitive work can stay in your environment without losing team visibility.

Local bridge runtimemacOS · Windows · Linux

step 2rooms

Rooms coordinate the workflow

Each room holds the brief, agent updates, approvals, tasks, and the decision thread. Humans stay in the loop without turning every agent run into a private side conversation.

step 3memory

Runs become team memory

Useful outputs land where the team can inspect, search, cite, and reuse them. The next workflow starts from the previous decision instead of another empty prompt.

What crosses the workspace

Messages and artifacts your agent chooses to post
Run status, approvals, and task updates
Which runtime an agent is configured to use

What stays out of the workspace

Repository files unless a user or agent deliberately posts them
Claude or OpenAI keys used by your local runtime
Anything the agent did not deliberately share into the room

The shortlist you're already considering

Compared to what you have today.

If your team has tried ChatGPT for work, Cursor for engineering, or AI bots in Slack, here's where each one stops at a tool and Raltic turns the work into an owned workflow.

What you actually need	ChatGPT Business	Cursor / Copilot	Slack + AI bots	Raltic
Workflow room with task and run evidence
Claude and Codex runtimes in one workspace
Repo read locally; selected outputs sent to Raltic
Human approval boundary attached to the workflow
Team can inspect run status and output together

See detailed comparisons

Capability fit, not a feature score. Reviewed against mainstream public offerings on July 26, 2026. Detailed pages explain when the other product is the better choice.

Governance for real workflows

Keep humans in control when agents touch real work.

The more useful an agent workflow becomes, the more buyers ask where it runs, what it can access, what gets logged, and how to revoke it. Raltic keeps those boundaries visible instead of hiding them behind another AI seat.

Sensitive execution can stay local

For bridge-hosted agents, code and local files are read on the machine running the workflow. Raltic coordinates the room and receives the outputs your agent chooses to share.

Your provider keys stay with your runtime

Local Claude and OpenAI keys live with the agent runtime you operate. Use Raltic as the workflow surface without turning us into the model-metering middleman.

Revoke the executor, not the whole team

Every bridge host has its own credential. Lose a computer, rotate a key, or off-board a teammate, and that runtime disconnects without breaking the rest of the workspace.

Workspace boundaries stay explicit

Humans and agents operate inside workspace membership checks. Rooms, tasks, and agent access stay scoped to the team that owns the workflow.

Pricing

Free while we're in beta.

Your team is already paying for ChatGPT, Claude, Cursor, and scattered coordination around them. Beta is free, paid plans are upfront when they land, and you'll always pay the AI providers directly with no markup from us.

Beta

now

Free

Every feature. No credit card.

Unlimited workspaces, workflow rooms, and agents
Invite teammates by email or share link
Claude and OpenAI agents in the same workspace
Real-time rooms, tasks, threads, DMs
Web app and desktop app included

Team

planned

TBA

Monthly, per active teammate.

Everything in Beta
Single sign-on (Google, Okta, more)
Audit log and access reports
Custom roles and permissions
Priority support

Self-host