AI-APM
All insights
AI for PMs
Productivity
Prompting

How project managers should actually use AI in 2026

·8 min read

Most project managers have tried AI by now — usually ChatGPT for a status update or a meeting summary. The honest result: it saved ten minutes, the output was generic, and they quietly went back to the old way.

The teams getting real leverage from AI in 2026 aren't using better prompts. They're applying AI to the parts of the PM job where it genuinely compounds: drafting, synthesising, and reasoning over the documents and conversations that already pile up around every project.

Where AI actually helps a project manager

Skip the listicles. There are five repeatable wins. Everything else is marginal.

  • Status reports. Feed the AI your raw notes, the last status pack, and the RAID. Ask for a draft in your house style. Edit, don't rewrite.
  • RAID grooming. Paste a meeting transcript. Ask: "What risks, assumptions, issues or dependencies surfaced here that aren't already in this RAID log?" You'll catch two or three every time.
  • Stakeholder comms. Same message, three audiences: exec sponsor, working team, external vendor. AI is fast at register and tone shifts.
  • Document synthesis. Drop a 60-page requirements doc and ask for the 10 decisions a PM needs to track. This is the single biggest time saver.
  • Plan critique. "Here's the plan. Here's the team. Where am I likely to slip?" AI is a surprisingly good sceptical reviewer when you ask it to be one.

A prompting pattern that works for delivery work

Generic prompts give generic output. The pattern that consistently produces usable PM artifacts has four parts:

  1. Role. "You are a senior delivery lead reviewing a weekly status pack for a FTSE-100 transformation programme."
  2. Context. Paste the actual artifacts — last week's pack, the RAID, the latest standup notes. Don't summarise; paste.
  3. Task. One specific output. "Draft this week's status pack in the same structure."
  4. Constraints. "Max 400 words. Flag anything that has slipped vs last week. Don't invent metrics — say 'TBC' if missing."
Why constraints matter
AI will invent numbers if you don't tell it not to. Every PM who's been burned by an AI status report has been burned by hallucinated metrics. "Say TBC if missing" is the most important seven words in your prompt.

What not to ask AI to do (yet)

  • Make decisions. Use it to surface options and trade-offs. Owning the call is still the PM's job.
  • Estimate from scratch. AI has no idea how slow your procurement process is. Use it to challenge an estimate, not produce one.
  • Handle confidential data without guardrails. If your org hasn't approved a tool, don't paste contracts, salaries, or customer PII into a public model. Use your enterprise instance.

A 30-day plan to actually change how you work

  1. Week 1. Pick one artifact you produce weekly (status pack is best). Generate it with AI for four weeks. Keep a note of time saved.
  2. Week 2. Move to RAID grooming after every steering meeting. Transcript in, delta out.
  3. Week 3. Use AI as a plan reviewer. "Here's the plan. What's missing?" once per sprint.
  4. Week 4. Build a prompt library. Save the three prompts that worked. Share with your team.

That's it. Most PMs who follow this saw 4–6 hours back per week by week four — not from AI doing the job, but from AI removing the keyboard-time around the job.

Where to go from here

If you want a structured run at this with real templates and worked examples, our learning paths cover this end-to-end for PMs and PMOs. The course catalogue breaks the same material into shorter modules if you want to dip in.