How to Use AI to Run a Post-Project Review

Before the post-project review meeting, use AI to synthesise lessons from the project documentation. This gives the meeting a structured starting point rather than relying entirely on recall.

Workflow: gather the key project documents (project plan, status reports, issue log, decision log, key communications). Upload to Claude or ChatGPT and ask:

'Review these project documents and identify: (1) the three things that worked well that we should repeat in future projects; (2) the three things that created problems that we should change; (3) the moments where the project was at most risk and why; (4) any patterns in how issues were identified and resolved.'

This AI-generated analysis gives the review meeting a concrete starting point and ensures that lessons are grounded in documented evidence rather than selective memory.

The post-project review meeting benefits from a structured format that goes beyond general 'what went well / what went badly' discussion.

AI can help design the meeting structure. Ask: 'Design a 90-minute post-project review agenda for a [type of project] with a team of [size]. The meeting should identify specific, actionable lessons for future projects rather than just general observations. Include questions that probe root causes rather than just symptoms.'

For complex projects where participants may remember events differently, having the AI-synthesised documentary analysis available during the meeting provides an objective reference point. When a participant says 'the plan was never realistic,' the AI analysis of the project documentation can show when this concern was first raised, by whom, and how it was (or was not) addressed.

The most common failure mode in post-project reviews is converting observations into vague lessons that cannot be actioned. 'Better communication' and 'more realistic planning' appear in every review but change nothing.

Ask AI to help convert observations to actionable guidance: 'Here are the lessons identified in our post-project review: [paste observations]. For each lesson, what specific change to our process, toolkit, or governance would prevent the same issue recurring? Express each as a concrete change with an owner and a timeline.'

Also useful: 'Which of these lessons are specific to this project and which are likely to recur across future projects? For the ones likely to recur, what systemic change would address the root cause rather than just this instance?'

This conversion from observation to systemic recommendation is where most post-project reviews fail. AI cannot make the decision to change processes or assign ownership; it can structure the options and the logic clearly enough that decision-makers can act.

Individual post-project reviews are only valuable if their lessons are accessible and applied to future projects. AI can help build and maintain the institutional knowledge that makes review investment compound over time.

After the review, ask AI to synthesise the lessons into a standard format suitable for a lessons-learned library: 'Summarise the lessons from this review in a standard format: project type, project size, key lessons, recommended changes, and any templates or tools that should be developed or updated.'

Store the synthesis in a searchable location (SharePoint, a knowledge base) where it can be retrieved before future similar projects. Before a major new project, ask Copilot or Claude: 'What do our lessons-learned records tell us about the common issues in projects like [this project type], and what should we do differently as a result?'

This use of AI to make lessons-learned accessible and applicable to future projects is where the investment in post-project reviews pays the highest returns.