Raq.com brings 70+ business and AI tools into one account. Same files, same boards, same record — whether the work is done by a person or an agent.
To do
Doing
Needs approval
Shared record
People see what agents did. Agents continue what people started.
Approval controlled
Sensitive actions pause for human sign-off.
01 — Why Raq exists
Accumulation is not a system. Raq replaces a pile of logins with continuity between tools.
Small and medium businesses add a chat, a board, a signing service, a drive, a form builder, a monitor and a handful of AI subscriptions. Each has its own login, billing and audit trail. An agent that can see only one piece is less useful.
Raq puts those jobs in one place with a common route for agents. A task can start with a message, pull a file from File Drop, create a document, request approval and leave the result where the team can inspect it. The value is breadth, shared data and the agent layer across the catalogue — not a claim to replace every specialist.
Shared
One account holds users, data and permissions
Switch accounts if you run several companies or client workspaces. AI usage and limits stay scoped.
Visible
Browser and agent work against the same records
A person can amend what an agent made. An agent can continue what a person started.
Controlled
Access is account-scoped with approvals where it matters
Tool access can be limited. Voice Calls and Ask My Agent use explicit approval stages.
02 — Catalogue
Every paid plan includes the public catalogue. Higher plans provide more AI usage and higher service limits. Some tools suit more than one group — listed once for clarity.
Personas and skills are reusable across the account.
03 — One workspace for people and agents
Raq exposes its tools to compatible agents through MCP. An agent does not need to imitate clicks when a structured operation exists: list boards, create a card, run an image batch, read a project record.
Supercomputer — built in
Describe the job in plain English. It chooses and runs suitable Raq tools with the same files and team.
Bring your own agent
Connect Codex, Claude Code or another MCP client. One connection reaches the account — scoped and auditable.
Supervise and correct
Orchestrator, Nature Work and Agent Operator HQ handle templated runs, remote terminal views and multi-terminal command.
People
Browser
Sees and amends
Raq account
Shared
records
Tasks · files · approvals · messages · audit
Agents
MCP operations
Structured — not clicks
Access is account-scoped. Sensitive or external actions use explicit approvals. Ask My Agent supports two-stage approval; Voice Calls is approval-controlled.
Voice Calls
From Professional — recorded & transcribed
Ask My Agent
Structured human ↔ agent with approvals
04 — How the tools work together
05 — Extend Raq
One route is self-built and sandboxed. The other is a managed product. Responsibility, runtime and support are intentionally different.
Custom Modules
First mock-up normally within one working day, working module within five working days after approval. Larger jobs flagged at mock-up.
£190 / $259 per module / month on top of plan
Your coding agent builds an application that runs inside Raq.com in an isolated container.
Best when you can direct and maintain the build. Raq provides the runtime and account connection.
Raq and Vu design, build, host and maintain a native module for your account.
One module = one area of work. Second module = second fee; modules can share data. AI uses normal allowance.
06 — Pricing
Per-user subscriptions. All public tools included. Higher plans raise AI usage and service limits. 5% off with annual billing. Prices exclude VAT.
Starter
£14 / $19 per user / mo
One use of each AI tool in trial; monitoring limits. Website Build is per-site add-on at this tier.
Professional
£45 / $61 per user / mo
Voice Calls starts here. More AI usage; higher monitoring & FinUK limits.
Business
£200 / $272 per user / mo
High AI allowance and service limits for teams running agent ops at volume.
Enterprise
£500 / $680 per user / mo
Highest limits, enterprise review and the full agent operations stack.
07 — Privacy & security
How AI privacy works
Core text and chat requests use Zero Data Retention routes by default — the provider cannot keep prompts or responses for training. You can change the setting to reach models without this protection. Image, video and voice tools send prompts/media to specialist providers under their own retention terms.
Raq does not train models on customer content.
Controls in place
See Privacy & Subprocessors on raq.com for current detail.
Start free — no card required
One subscription for the catalogue. Use the browser, bring your own agent, or let Supercomputer run the job — all on the same account.
Built by Vu · Suite 15, 1666 High Street, Knowle, Solihull, B93 0LY · hello@raq.com