Practical AI for MSPs

Let’s cook up something useful.

Field-tested recipes, honest lessons, and practical playbooks for building AI inside a modern managed service provider.

The recipe box

Pick what you’re trying to make.

Short on theory. Heavy on things you can try with your team this week.

Recipe No. 002Build

Design an AI intake process your team will actually follow

Turn scattered ideas into a focused backlog of useful, measurable AI work.

Recipe No. 003Operate

Give your service desk an AI sous-chef

Start with the repetitive work, preserve human judgment, and measure what changes.

Recipe No. 004Sell

Package AI readiness as an MSP service

A simple offer that helps clients move from curiosity to a responsible first project.

Before you start

AI mise en place.

Three ingredients we keep on the counter for every AI project.

  1. 1

    Begin with real work

    Find the frustrating workflow before reaching for the newest model.

  2. 2

    Keep a human in the kitchen

    Build judgment, review, and accountability into the way the tool is used.

  3. 3

    Measure the meal

    Decide what better looks like, then check whether you actually made it.

Fresh from the kitchen

Field notes.

The honest version of building AI at an MSP—decisions, mistakes, and what held up in production.

01

From the field

The demo is not the product

Aug 20264 min
02

Strong opinion

Your AI policy should help people say yes

Aug 20266 min
03

What I learned

Start with the workflow, not the model

Jul 20265 min

Why Cooking Up AI?

We’re building this in the real world, not a lab.

We’re Anthony Latham and Graham Rosenberg—two Sourcepass operators working through the practical questions that appear when AI moves from slide decks into service delivery.

This is where we share the playbooks, patterns, failures, and strong opinions we wish we had when we started. Everything here is for people with clients to serve, teams to support, and no appetite for empty hype.

Meet the people in the kitchen