AI and Kanban in MSP project management: what each one is for

MSPs ask whether they should manage projects with AI or with Kanban. They do different jobs, and they work best together.

A Kanban board is where the team decides what to work on and pulls it. It shows every ticket and project task in one place, limits how much is in progress, and gives the morning huddle (a 10 to 15 minute meeting at the board) something to look at. People make the calls.

AI is good at reading a lot of records quickly and writing up what it finds: the notes on a project nobody has opened in a month, the time entries that don’t add up, the new tickets that need a sensible subject and a type. That is work a service manager or project manager rarely has time for.

What AI does well for an MSP

• Reading notes and time entries. It can summarize what happened on a project since Monday, or what a tech had open before they went on leave.
• Spotting projects that drift. A project marked Active with no hours logged in a month, or 85% complete with half its tickets open, is easy for a model to flag and easy for a person to miss.
• Triage and cleanup. Setting the type, subtype and item on new tickets, or rewriting a subject so it says what is wrong.
• Drafting. A status update for a client, a handover note, a quarterly business review (QBR) summary from the open tickets.

What it doesn’t replace

AI doesn’t decide what matters most this week, and it doesn’t make a team finish before starting. Those still come from the practices behind the board: a ranked queue, techs pulling work, work-in-progress (WIP) limits, and a daily huddle. The Kanban Method for MSPs covers them.

How TopLeft uses AI today

Project Health AI (early access)

Project Health scores every project in ConnectWise, Autotask or HaloPSA from the data TopLeft already syncs. Six checks feed the score: schedule, budget, activity, completion, margin and milestones. An AI review then reads the raw facts, including the notes and time entries your team writes, looks for places where they disagree, and says which projects need attention and what to do. A portfolio review lists the projects that most need action and three to five actions for the week.

The review runs on your choice of OpenAI or Anthropic, with zero data retention. You can opt any client out, or anonymize client, project and people names so no free text is sent at all. Project Health is available on request while it is in early access.

The MCP connector (beta)

The MCP connector (MCP, Model Context Protocol, is the open standard AI tools use to connect to other software) lets the AI tool your team already uses, such as Claude, ChatGPT, Gemini, Copilot or n8n, work with TopLeft. You can ask it which tickets haven’t been touched in 14 days, what a tech was working on, or what is open for a client. It can also create projects, tickets and tasks, triage new tickets, add notes and log time, and every change is written back to your PSA the same way an edit on the board is.

TopLeft doesn’t send your data to an AI provider for the connector. Your AI tool asks for it, under your account and your AI policies, and it sees only what you can see in TopLeft. Your admin decides who can connect and whether their AI can make changes.

When to use which

Ask the AI when:

• a client wants a status and you don’t want to open ten tickets,
• a tech is out and you need to know what they had open,
• new tickets need a subject that makes sense and the right type.

Open the board when:

• you’re planning the week and moving work around,
• you need to see everyone’s workload side by side,
• you’re running the morning huddle.

To see the board, Project Health and the connector on a demo environment, book a demo.