AI customer zero.

MSPs who plan to learn AI and adopt it internally before offering it to clients have a timing problem.

Going first makes sense. A small AI pilot starts with the most senior engineers or the devops team, and what they learn guides the offer. A pilot that stops at teaching engineers a tool leaves the business as it was. One aimed at a real problem in the MSP's own operations improves the business first, and gives the client offer something that has already worked.

Until the pilot finishes, there is no offer for the clients already asking for help. Those clients either adopt AI themselves without guardrails or turn to a competitor. Traditional MSP services are a red ocean. AI advice is still blue water, and the MSPs that get there first can grow in it.

The pilot group is also keeping the lights on

Escalations, projects and keeping the lights on (KTLO) still land on those same engineers, who are contested resources juggling multiple priorities. Their time goes to whatever is on fire, so the pilot slips by a day, then a week, then a quarter unless dedicated time is carved out for it. Any project that shares people with day-to-day operations follows the same path. Article 3 of this series, 3 clocks, 1 plan, covers why a decision lands at a different speed for the owner, the leadership team and staff.

Capability built around KTLO runs on its own clock.

Going first needs a project around it

The plan to go first holds up with 3 things in place: someone who answers for it, hours to do it and an agreed end.

A named owner with an executive sponsor

The owner answers for the result. The sponsor is the senior person who backs the owner's decisions when a client escalation competes with the pilot for the same engineers. An owner without a sponsor can ask for time but cannot protect it. A sponsor without an owner has approved a pilot that no one runs. With either missing, the pilot's priority is low.

A budget with dedicated hours

In practice this is an approved project with an approved time allocation, run like any other project. The allocation says how many hours, from whom, and until when, and it includes the hours for discovery. Engineers log their time on timesheets and go through the normal review and approval process. Licences and tools get approved through the procurement process, and when the pilot's allocated time is not protected, the day job wins.

A definition of done

This is the scope of the pilot and the capability uplift it is meant to deliver. Scope starts with discovery.

The pilot group finds the bottleneck it is meant to relieve in the MSP's own operations, the place where work waits. It observes how that work runs day to day, including how exceptions get handled, documents what it sees and validates the result with the people who do the work. What it finds can differ from the process document. Design follows discovery.

Learning has no natural end point, so a pilot without one runs until something more urgent takes it over. Left undefined, "still learning" is always a reasonable status, and the offer to clients never gets a launch date.

An internal pilot needs someone accountable, a budget behind it and an agreed outcome.

Aptimus works alongside the people running the pilot until it has an owner, a budget and a definition of done. If your MSP is already testing AI on itself and clients are asking, aptimus.com.au is the right place to start.