About

Built by an MSP.
For MSPs.

It didn't come out of a lab. It came off our own service desk. Same ConnectWise stack. Same tickets. Same risk of letting AI loose on a client's systems.

How we got here

We're a ConnectWise MSP, and we've spent years building for our customers. Like everyone, we started with Copilot. Then we went all-in on ChatGPT for a few months. Then Claude. Then back and forth between the two giants, chasing whichever was best that quarter, until we realized the back-and-forth itself was the problem. We didn't want our operations to depend on one vendor's roadmap, pricing, or mood.

So we went deeper than was reasonable. We wrote ourselves on AI, from "what is an LLM" to production LLMOps, because every time we trusted the hype instead of our own notes, we got burned. We stopped reading vendor benchmarks and built private eval sets on our own real tickets and workflows, because the only leaderboard that predicts anything is one built on your own tasks. We audit our own AI configurations with the same rigor we audit a client's network, and we keep a written record of every mistake we've made along the way. Thirty-two lessons and counting.

Pulsatrix is what all of that became.

What we kept seeing

Talking with partners, colleagues and friends, the same pattern came up everywhere. Everyone wants to save time and automate everything. But most of them aren't software engineers, they already have a full-time job. They're riding the hype and quietly losing the hours: babysitting half-finished automations, hoping the payoff finally comes, more tired than they were before.

And if you've been vibe-coding your way to "almost done", honest questions: who ships it to production? Who maintains it? Who notices when something breaks at 2 a.m.? Who's watching for unauthorized access? Who owns the API keys, the PII, the model bill? Right now, the answer is you. On top of the job you already had.

LLMs save time only when they are structured. Without governance, “AI” is just a new layer of work.

We built Pulsatrix against that trap. Pulsatrix exists to empower technicians, not replace them, and not to quietly turn them into unpaid platform engineers either.

How we work

Principles we won't trade away

The same standards we use on our own desk. Applied to every MSP who runs Pulsatrix.

Empower, don't replace

We make the tech in the seat better, not redundant. The people are the point; the tool just lifts what they can do.

A tool, not autopilot

It works while you work and acts only when you do. Nothing runs unattended on a client's systems, a smaller attack surface, by design.

Bring your own model

No new AI to adopt or trust. Point the subscription (or the local model) you already pay for at ConnectWise.

Runs on your turf

Self-hosted on your own infrastructure. Your clients' information never has to leave your office.

The path so far

A mature MSP. Not a startup story.

Our MSP · roots
Decades in the field
Our MSP company is long-running. Roughly thirty years of serving real customers, not a lab pitch. Production desks, production SLAs, production pain. That is the company that built Pulsatrix.
Our MSP · stack
ConnectWise in the blood
Same PSA, same tickets, same risk of putting AI next to a client's systems. We learned AI on our own board (Copilot, ChatGPT, Claude, local models, evals on real work) before we shipped a product to anyone else.
Learning · the hard way
Notes, evals, sovereignty
A deep internal knowledge base, private evals on real tickets, open models on our own Mac. We stopped trusting vendor scoreboards and earned the right to talk about local-first in production.
Now · early access
Shaped with the MSPs who run it
Application-based access. Plans and depth shaped with the desks that live in ConnectWise every day. Feedback is the product, write us when we get it wrong.

Local-first · ConnectWise · Human approval

Try it

The feedback is the point.

Run your ConnectWise stack in plain language, keep the key, and tell us what we got wrong.

Talk to us

Built by an MSP.For MSPs.