MSPs right now are staring at three options. Build their own AI tools in-house using open-source platforms. Buy from one of the dozen VC-backed AI startups pitching free trials in your inbox. Or wait for ConnectWise, N-able, Kaseya, or Datto to bake AI into the PSA they already pay for.

I’ll tell you upfront: only one of these is a real strategy for most MSPs. But which one depends on your size, your team, and how honest you’re willing to be about both.

Option 1: Build In-House

The build camp got louder over the past six months. Channel Dive talked to MSP owners who are using open-source agentic platforms to spin up internal AI tools — help desk automation, documentation workflows, custom integrations — without dedicated dev teams. KeyStone Solutions CEO Preston West put it plainly: “I think there is definitely something there around bespoke development and automation and integrations that people like us can do without a huge, dedicated dev team, which it would have taken even just a couple of years ago.”

That’s the pitch. And it’s real. The tools are genuinely more accessible than they were 18 months ago. If you have one person on your team who can vibe-code a workflow and isn’t afraid to experiment, you can ship internal tools that would have required a software contract a few years back.

The honest problem: you’re running an MSP, not a software company. OpenText Cybersecurity VP of Business Development Mike DePalma said what every PSA vendor is thinking but won’t say to your face: “Do you want to be an AI vendor, or do you want to be an MSP? You can, but to be able to keep up with the changes — that’d be your full-time job.”

He’s right. Building in-house means owning the maintenance, the updates, the security review, and the opportunity cost every time you’re debugging a workflow instead of closing a deal. For MSPs under $5M ARR without dedicated technical staff, this path is slower than it looks from the outside.

For MSPs with PE backing or a dev-capable team? It’s a real option, and it’s producing real results for the firms that have the resources to do it right. The PE-backed MSP churn and AI headcount trap is worth reading if you’re in that category — the pressure to automate comes with its own set of problems.

Option 2: Buy From Niche Vendors

The niche AI vendor market is chaotic. There are good products in here. There are also a lot of “this demo was built last Tuesday” startups that will fold in 18 months.

The integration question is the one that kills deals. Most of these tools promise to connect with your PSA. Some of them actually do. The ones that don’t leave you with an AI tool that runs parallel to your existing stack instead of inside it — which means double data entry and a workflow nobody follows after the first week.

What’s working: help desk AI that triages tickets and suggests resolutions inside your existing PSA interface. What’s almost working: documentation AI that pulls from your RMM and creates runbooks with one click. What’s mostly marketing: anything claiming to “fully automate” tier-1 support end-to-end.

West’s framework from Channel Dive is the right one here: rank your top AI priorities that will actually move customer outcomes, and only buy tools that address those specifically. “We can’t always be chasing the better mousetrap, because we’ll never actually get anything done,” he said. Sound advice from someone who’s been in the fire-hose moment.

The other filter: check whether the startup has an acquisition path. N-able CTO Mike Adler told Channel Dive that N-able is actively tracking AI startups for acquisition, looking for ones with a genuine “technical moat” — proprietary data, a differentiated model, something that can’t be replicated in six months. If the startup you’re evaluating doesn’t have that, the PSA will build or buy something similar in 12-18 months and undercut the license fee.

Bet on the tools with moats. Pass on the feature wrappers. The Kaseya vs. ConnectWise vs. Pax8 platform wars are the competitive context here — whichever platform you’re on shapes which AI bets make sense.

Option 3: Wait for Your PSA

This is where most MSPs actually are, and it’s the most defensible position if you’re honest about your bandwidth.

The PSA case isn’t just patience — it’s data. N-able, ConnectWise, and Kaseya collectively manage millions of endpoints. Omdia MSP Practice Leader Robin Ody made the point that this data advantage is real: “This data is important because it allows the vendor to sift masses of real language data to build non-deterministic AI that can take the often incomplete ticket data for example, and make decisions on its own.” A niche startup can’t replicate that corpus without years of deployment.

I tested what’s actually live in PSA and RMM platforms earlier this year. Most of the AI that’s generally available today is incremental — smarter autocomplete, ticket routing improvements, better search. The agentic workflows that PSA vendors are promising — where the tool handles an entire tier-1 resolution chain without human intervention — are in beta or roadmap. Not in your queue today.

The PSA wait is the right call if you need stability over speed. The risk is what DePalma said: “What we’re doing now, six months from now, is going to look kind of ancient.” If you’re waiting for ConnectWise or N-able to catch the niche vendors, you might be waiting 18 months while the fast movers pull ahead on operational efficiency.

The Real Framework

Here’s how to decide. Ask three questions:

Do you have someone on staff who can own AI development? Not a tech enthusiast who dabbles. Someone who will spend 20+ hours a week on it. If yes, build-or-buy from niche vendors is viable. If no, wait for your PSA and use that time to document your processes so you’re ready to automate them.

Is your AI goal internal efficiency or a client-facing product? Internal efficiency is where PSA vendors are closest to ready. Client-facing AI products — where you’re reselling or white-labeling AI tools — are where niche vendors and in-house builds have the edge today.

Are you PE-backed? If yes, you have resources and pressure to move faster. Build or buy aggressively, with a portfolio lens on which tools become strategic assets and which are commodity purchases your PSA will replicate.

The MSPs who are going to win this aren’t the ones who pick the right tool. They’re the ones who make a decision and go deep on it. West’s warning is the one to remember: the firms that wait for the perfect answer will watch everyone else pull ahead. “If you sit there and you really nitpick and you’re waiting for this one feature, everybody’s going to catch up.”

Pick your path. Go deep. Stop watching the demo reel. The MSP bifurcation growth trap is what happens when you don’t — the gap between the haves and have-nots is widening every quarter.