MSP automation skills gap: can AI workflow builders actually close it?

The gap by the numbers

MSPs agree on the goal. In Rewst's 2026 State of MSP Automation report, 95% of MSPs said automation is important or essential to their strategy, and expertise was named as a primary barrier to scaling it. The supply side is thinner than the demand. Speaking at Rewst's FLOW 2026 conference, CEO Aharon Chernin estimated there are about 80,000 MSPs worldwide and about 1,000 automation engineers able to build automations for them. Rewst itself works directly with about 1,500 MSPs. His question to the room, as ChannelE2E reported it: "What do the other 79,000 MSPs do?"

These are a vendor's estimates and a vendor's usage data, so treat them as directional. Rewst reports its customers ran more than 19.5 billion automated tasks over the past year, nearly 2.4 times the prior year, and the company does not publish a definition of what counts as a task. Even so, the direction matches what MSP owners describe: automation demand is compounding while the pool of people who can build it is not.

The industry answer: let the agent draft the workflow

The answer taking shape across the channel is the AI workflow builder. Rewst announced AI-native capabilities in June, built around an agent where a user describes a problem and the agent builds the workflow on a canvas the user can view, edit, and approve. ChannelE2E reported that Rewst re-engineered the product around its MCP server for this, tested it with about 30 MSPs, and positioned it so that non-technical staff can work with it much as they already prompt a chatbot. On October 16, ChannelPro's weekly roundup listed the platform as globally released.

It is not just Rewst. The same roundup lists LogMeIn Resolve adding an MCP server, Alternative Payments adding agentic accounts receivable and payable with human approval before actions, and Blancco adding a plain-language agent that connects to asset-management tools through MCP. The pattern is consistent: describe the outcome in plain language, let an agent assemble the steps, keep a human approval gate. For the governance side of that pattern, see the MSP AI governance gap around MCP.

The business case is already showing up in MSP plans. In July, ChannelE2E reported that Matson and Isom Technology Consulting intended to take these capabilities to SMB clients by October and make them central to its 2027 business plan, while Summit IT Solutions was exploring Automation-as-a-Service for its own SMB clients. Chris Harp, M&I's COO, framed it as something that can "elevate, not eliminate positions."

What an AI builder genuinely fixes

It fixes the blank-canvas problem. A technician who knows exactly how onboarding should work, but cannot write the logic, can now produce a first draft. Chernin's claim, as quoted by ChannelE2E, is that engineers can "build 10 automations a week instead of one." Martin Perkins of Tech Cartographer told the same outlet that the tool "really lifts the weight of needing a full team devoted to this." Those are real gains for a ten-person MSP that has never had an automation hire.

It also changes who is in the loop. The Rewst partner quote in the October roundup is telling. John Snyder, CEO of Net Friends, said staff can use the agent to draft a workflow "and hand it to our integration engineers." The agent shortens the front of the pipeline. A human with integration skill still sits at the back of it.

What it leaves open: the integrations underneath

Every workflow an agent drafts calls into other systems: a PSA, a remote monitoring tool, an identity provider, an accounting package. Alex Williams of M&I made the point directly when describing client use cases, saying you can interface through the API and set up guardrails so a client can safely interact with QuickBooks. The guardrail is only as good as the connection beneath it, and that connection belongs to a third party that changes its API, its field names, and its permission model on its own schedule.

This is where the skills gap quietly reappears. Writing a workflow becomes cheap; verifying that the thirtieth workflow still maps the right field after a vendor update does not. An MSP that multiplies automations across dozens of clients multiplies the surface area where a silent break can occur, and a drifting field mapping tends to surface as a wrong invoice or a missed offboarding rather than a clean error. That is the problem our self-healing integration approach targets: detect when an upstream change breaks a mapping, then re-map and repair the connector so the workflow above it keeps telling the truth.

Packaging it: from tool to service line

The channel is also building the commercial rails. Pax8 told Channel Dive in June that it is adding token tracking to its MSP operations platform, with COO Craig Donovan saying "usage is probably going to be the predominant monetization strategy in the agentic world." Pax8 also simplified billing for tools including Rewst's RoboRewsty. If automation is going to be sold on usage or as a managed monthly service, an MSP's margin depends on the cost of keeping each delivered automation running, not the cost of building it once.

That argues for treating integration upkeep as a shared layer rather than a per-client chore. The model we describe in offering integration as a managed service is built on this idea: the MSP owns the client relationship and the billable outcome, while the connection layer is monitored and repaired centrally. Our partner program is designed for that split.

What MSP owners can do this quarter

First, inventory the integrations your existing automations depend on, and note which vendors have shipped API changes in the past year. Second, decide who owns repair: a named engineer, a vendor, or a platform. Third, if you are piloting an AI builder, require that every approved workflow lists the systems and fields it touches, so a break can be traced. None of this slows adoption. It prevents the first wave of AI-built automations from becoming the next backlog.