Endpoint documentation built for MSPs
Run many clients from one place. Ambiscribe keeps a current, per-client picture of every endpoint and a field-level history of what changed, so your techs and their AI agents stop rediscovering each environment on every ticket. Not another RMM.
Ambiscribe is a multi-tenant documentation and change-detection layer for MSPs. Each client's fleet is documented automatically every five minutes and scoped to its own view, baselines, and retention. It complements your RMM and PSA. It isn't one, and it doesn't act on machines.
The problem
Every ticket starts with rediscovery
A ticket comes in for a client you last touched a month ago. What's installed, what changed recently, whether the security posture slipped, who has admin. For a tech that's friction on every ticket, repeated across every client you run.
For the AI agent you'd like to hand first-line triage to, it's worse. Without structured access to that client's current and historical state, the agent is guessing, and a guess isn't something you let it act on.
Ambiscribe holds that state for every client, already collected and already diffed, scoped so a tech or an agent only ever sees the client they're working.
Built multi-tenant
How it fits an MSP
Every page scopes to a client from one dropdown. Baselines, retention, and access all follow the client boundary.
Multi-tenant by client
One dropdown scopes every page to a client. Built for running many fleets at once, not one big pool you filter.
Per-client baselines
Define the expected state for each client and see a color-coded pass/fail matrix the moment a machine drifts.
Retention & contacts
Set history retention and the people to reach per client, so each one is configured to its own contract.
One-line enrollment
Generate a platform-specific install command per client. The device auto-assigns on its first report.
Scoped roles & keys
An msp_admin sees all org clients; scoped roles see only theirs. Two-factor sign-in and scoped API keys for services.
Notable-change push
HMAC-signed webhooks and email on notable changes, ready to feed your PSA and the documentation tools you already run.
AI agents per client
The MCP tools are scoped per client, so a client's agent queries that client's state and nothing else.
Network devices included
A LAN probe documents each client's firewalls, switches, and access points in the same fleet view.
Org-scoped audit log
Actions are recorded and attributed, scoped to your organization, so you can answer who changed what.
Where it fits
It complements your stack, it doesn't replace it
Your RMM acts on machines. Your PSA runs the tickets and billing. Your wiki holds the procedures and the context behind a setup. Ambiscribe is the layer underneath all of it: the always-current record of what each client's fleet actually is, and what changed.
It pushes configs, inventory, and notable changes into the tools you already run, and it stays in its lane. It documents and answers. It doesn't run commands, push patches, or take action on endpoints, which is what keeps it safe to point an AI agent at.
Questions
Common questions
Is Ambiscribe an RMM?
No. It's a documentation and change-detection layer. It records each client's endpoint state and what changed, and it doesn't run commands, push patches, or take action on machines. It complements your RMM and PSA rather than replacing them.
How does multi-tenancy work for MSPs?
Every page scopes to a client from one dropdown, with per-client baselines, retention, and contacts. An msp_admin sees all of the organization's clients; scoped roles see only their assigned ones. Device-to-client assignment is driven by the agent's enrollment key, so re-enrolling under a different client moves the device and keeps its history.
How do I onboard a new client?
Generate a platform-specific enrollment one-liner from the Clients page. It bakes in the server URL and key, installs the agent on Windows, macOS, or Linux, and the device auto-assigns to that client on its first report.
Can I give a client's AI agent access to only their data?
Yes. The MCP tools are scoped per client, so an agent queries one client's current and historical state without seeing another's. Engineers get the same scoping in the dashboard.
Stop rediscovering your clients' fleets
Give every tech, and every AI agent, a current per-client picture of what's there and what changed.
Request early access