Vendor-agnostic MSP security & compliance platform
One view of every device on every client network.
MSPs run each client through a scatter of disconnected tools — one for monitoring, another for compliance evidence, another for assets, another for identity hygiene. KEEP replaces the scatter with a single platform and a continuous, defensible record of what was found and what was done about it.
OPEN INCIDENTS: UPS_ON_BATTERY (raised 4 min ago, 1 log entry)
UPS STATUS: battery=41%, runtime=18min remaining
OTHER DEVICES UNREACHABLE: None — this device appears to be the only one affected.
Most likely: utility power loss to the server-room UPS. Battery is discharging under load with no other device on this client currently affected, which rules out a wider circuit failure.
Illustrative example of a Merlin AI device diagnosis, not a live result. Device Diagnosis is fully implemented and available today, still gathering real-world validation — see Capabilities for current status. (Source: Merlin architecture documentation)
Architecture
Deterministic first. AI where it earns its place.
KEEP is built in three layers. Only one of them depends on an AI provider being reachable — and nothing else in the platform notices if it isn't. (Source: KEEP architecture documentation)
The deterministic platform: monitoring, compliance evidence, incident management, asset tracking, identity, environmental and power monitoring, and reporting. Runs completely on its own — no AI dependency anywhere in onboarding, deployment, monitoring, or the audit trail.
The governance layer built into KEEP: policy enforcement, audit ownership, and rule-boundary checks that decide what any future automation is and isn't allowed to do. Ships with every install, at no additional cost, with no AI dependency at evaluation time.
The narrative layer on top: reads what KEEP already gathered and explains it in plain language. Today that means a ranked device diagnosis and coordination guidance on an active incident — both fully built, still gathering real-world validation.
A prioritized brief after an onboarding scan is also built, though it's not part of the current evaluation focus. On every surface, Merlin proposes; a person decides. (Source: Merlin architecture documentation)
What KEEP watches
The instruments an MSP actually needs in one place.
Compliance & monitoring
A health score per client, cross-client pattern detection, and a full incident command view the moment something goes critical.
Assets & identity
Hardware register, IT dependencies, warranty-versus-end-of-life gaps, and identity conflicts flagged before they become findings.
UPS & power
UPS runtime and battery-degradation trending, watched continuously today. Temperature monitoring is planned — scaffolded but not yet collecting from a real sensor — see Capabilities.
Network monitor
Critical infrastructure — firewalls, core switches, domain controllers, UPS — with plain-English device and incident status for every client.
Reporting
Site assessments, quarterly compliance, and EOL exposure reports, generated from real data, not assembled by hand.
Decommissions & records
A signed chain of custody from first discovery through disposal, so "what happened to this device" always has an answer.
Merlin
Merlin explains. It never decides.
Merlin is not a chatbot bolted onto KEEP. It's the same governed identity everywhere it appears, held to the same rules on every surface.
- 01No standing authority. Merlin only ever acts within what the technician or Director using it is already permitted to do.
- 02Propose, then a person approves. Every action Merlin suggests is confirmed by a human before anything happens — no exceptions in Version 1.
- 03Provider-independent. Which AI model answers is a configuration choice, not part of what Merlin is. Bring your own provider account, or use KEEP's, at the same price either way.
(Source: Merlin architecture documentation)
Device Diagnosis
Evaluation CapabilityReads open incidents, recent work log entries, event history, and UPS or temperature state for a device, then ranks the most likely explanations with the evidence behind each one.
Incident Coordination
Evaluation CapabilityReads the full incident thread across every technician working it and suggests what's already been tried and what to do next — logged permanently to the incident record.
Onboarding Scan Analysis
Evaluation CapabilityReads every completed onboarding scan for a new client — the firewall and every discovered device — and correlates the findings into a four-part remediation brief.
Built and available today; deprioritized relative to Device Diagnosis and Incident Coordination for the current evaluation focus.
Trust
No lock-in, on the platform or on the AI behind it.
No open-source licensing blockerwas identified anywhere in KEEP's own codebase during a direct, package-by-package audit of every dependency. (Source: KEEP architecture documentation)
Bring your own AI provider account, or use KEEP's — same price either way. The model behind Merlin is a configuration choice, not a dependency. (Source: Merlin architecture documentation)
Device inventory, incidents, and credentials stay on your own Hub, always. When Merlin's narrative layer is on, it sees only the descriptive context one request needs — never a credential, key, or token, by construction. (Source: Merlin architecture documentation)
Runs at a client site or aggregated across an MSP's whole book of business — one platform, not a different product per deployment shape. (Source: KEEP architecture documentation)
Evaluation Program
From first look to a working deployment.
- 01
Request
Tell us about your book of business. Evaluation access is still provisioned directly, not self-serve — though you can create a KEEP account yourself any time.
- 02
Deploy
Stand up a Hub for your MSP or a Spoke at a client site — the same install, the same platform, either way.
- 03
Evaluate
Full platform access from the start. Merlin's AI-dependent features need a provider key configured before they'll respond — everything else needs no separate setup step.
- 04
Decide
Move to a paid subscription once your evaluation term ends. No pressure while you're still finding out if it's a fit.
(Source: KEEP Evaluation Program documentation)
Evaluate KEEP on your own network.
Staff-provisioned, no charge during evaluation, Merlin included from day one.
Know · Evaluate · Enforce · Protect