Connect your Cleeng data to your own AI tools, with control
The Cleeng MCP server lets you connect your Cleeng data to your compatible AI tools. You remain in control of who can connect, what they can access, and when that access is removed.
Cleeng provides the access layer. You bring the AI account.
The Cleeng MCP server is built and hosted by Cleeng as part of the Cleeng Service. It is an additional way to access Cleeng data, alongside the Cleeng dashboard and API, not a separate third-party data service.
When you connect the server to Claude or another compatible AI tool:
- You connect your own AI account to your Cleeng account.
- You authenticate using your existing Cleeng credentials and permissions.
- The AI tool can request Cleeng data only on behalf of the authorised user.
- You remain in control of whether to establish, maintain or revoke that connection.
The foundation behind responsible AI
Cleeng brings subscriber data together through five governed layers, from raw feeds to the answers your teams and AI agents rely on. Each layer helps keep information accurate, consistent and traceable, so your data can be used confidently across Cleeng AI and your wider business.
Raw
Everything, exactly as it arrived. Immutable.
No silent fixes, every downstream number can be traced, audited and replayed to source.
Cleansed
Known-bad and out-of-scope records quarantined, visibly. Every exclusion labelled with the rule that made it.
Edge cases are handled explicitly, not left to quietly distort metrics.
Standardised
Every source translated into one vocabulary.
Channels become directly comparable before any business logic runs.
Modelled
Business logic composed: journeys, revenue, status history.
A definition lives in one place; change it once and it changes everywhere.
Marts
Consumption-ready, contract-stable outputs.
Dashboards, exports, shares and agents read from surfaces that don’t break downstream.
THE FIVE LAYERS · ~450 VERSIONED MODELS
Your data stays yours, at every level
Cleeng protects customer data through layered security controls across the environment, user access, shared resources and individual records. Every customer has an isolated data boundary, every person and service receives only the access they need, and permissions can be reviewed or revoked at any time.
Discipline that holds under audit, not just on a diagram.
People, services and agents each get the minimum they need.
Each partner sees exactly the slice granted, revocable at any time.
One governed model, many clients. No client can ever see another’s rows.
And the AI surfaces inherit it: agents operate inside the same permission model, including the no-PII reporting mode. The new interface gets no new privileges.
A practical checklist for your team
Before connecting
- Confirm the use case and business owner.
- Choose the minimum access required.
- Confirm that the user is authorised to access the relevant data.
- Review your AI provider’s terms, privacy settings and enterprise controls.
- Follow your organisation’s privacy, security and data-handling policies.
While using MCP
- Use only approved AI accounts and workflows.
- Review outputs before acting on them.
- Avoid sharing data beyond the approved use case.
- Report suspicious activity or unexpected outputs.
- Keep human oversight for material decisions.
When access changes
- Remove or update Cleeng roles when responsibilities change.
- Disconnect unused AI-tool connections.
- Reassess permissions and connected accounts regularly.
- Revoke access immediately if a credential or account may be compromised.
- Retain evidence required by your internal governance process.
Review the connector before you enable it
Send us your security questionnaire, or ask for the scope list and the function inventory. We can also open a sandbox account so your team can connect and inspect the responses before anything touches production.