About
We build for the MSP service desk, not the AI demo reel.
Ops AI is an automation runtime for Managed Service Providers, made by people who've worked the service desk. It runs in the background. It logs everything. It asks before it writes.
Why we built it
The problem we kept seeing.
Every MSP has the same shape: a PSA, an RMM, a documentation tool, a CSP portal, a security stack, and a service desk that stitches them together by hand. Engineers spend a huge share of their day doing work that's already written down — just not in a form a computer can run.
SOPs live in Confluence pages and Google Docs. At 2am, when the alert fires, the SOP isn't in the loop. An engineer reads the runbook, clicks through six tabs, and resolves the ticket. The runbook was good. The execution was expensive.
We built Ops AI because we got tired of watching that happen. Not because "AI will replace MSPs" — it won't, and anyone claiming otherwise is selling something. Because the patterns in service-desk work are legitimately automatable, if the runtime is designed for MSPs instead of retrofitted from a generic chatbot.
What we believe
The principles we build against.
MSP-native, not MSP-flavoured
The tools we ship speak Autotask, Datto RMM, IT Glue, Pax8, Microsoft CSP, UniFi, and Datto EDR — at their actual APIs, not at a generic "documents" abstraction. If a vendor ships a breaking change, our tests catch it. Because we only have seven vendors to care about, we care about each one properly.
Evidence beats adjectives
Every agent run shows you what happened, in what order, how long it took, what it cost, and why. No black-box magic. No "trust the model." If an agent made a decision, there's a trace you can read.
Approval-first
Every write action is gated unless you've explicitly said otherwise. AI that can't cause damage until you let it is a design choice, not a feature. The default posture is "propose, don't do" — and you can loosen it per SOP, per vendor, or per customer when you trust the pattern.
Boring in the places that matter
Microsoft Azure. Postgres. Entra. Stripe. Standard stack. We save the novelty for the agent runtime, where it earns its keep. Your infrastructure team doesn't need to learn anything new to run Ops AI.
Team
Small team. Operators first.
-
01
Operators first
Service-desk veterans before founders.
-
02
Audit-first
Every agent action traceable to a person.
-
03
Ship small, often
Customer-visible change every week.
-
04
Architecture in the open
Full data flow, retention, and isolation model on the Security page.
We're a small team. Founders and early engineers have run MSP service desks — the escalation rotas, the 2am RMM alerts, the licenses-about-to-expire-at-Christmas. We're building the runtime we wish we'd had.
We publish what's already true. The full security architecture — data-flow diagram, retention policy, tenant-isolation model — lives on the Security page. Pricing is open. We don't publish photos of a stock-image team; we'd rather you judge us on what we ship and on the traces every agent run produces.
Fit
When Ops AI isn't right for you.
Three honest disqualifiers. We'd rather you find this out before signing up than after.
You're a ConnectWise-first shop.
Ops AI is built for Autotask shops. ConnectWise PSA isn't on the integration list and we're not planning to make it the second flagship. If your service desk runs on ConnectWise Manage, you'll have a better time with ConnectWise Sidekick.
You don't have a Microsoft tenant.
Sign-in is Entra-only — there's no email-and-password option and no plan to add one. If you don't run on Microsoft 365 internally, you can't get into the product to evaluate it.
You want a no-code workflow canvas.
Ops AI is agent-led with a fixed set of vendor-specific tools — not a generic action canvas you wire together. If you want hundreds of integrations and a Zapier-shaped builder, Rewst is the closer fit.
Careers
Want to automate the parts you hated?
We're hiring engineers who've worked in MSPs and have strong opinions about what should and shouldn't be automated. Backend, frontend, and platform roles are open.
Not formally listing the roles yet. If the description above sounds like you, email careers@msp-ops.ai with what you've worked on and why this problem matters to you. Short and specific beats polished.
Talk to us
Product question, partnership, or just a hello?
We read every email and we reply within a business day.