Toolsday

September 29, 2026 · Last updated on September 12, 2026

AI Toolsday: Real AI, Real Agile, Real Workflows

# Continuous Improvement

Practical AI workflows for Agile teams

Find the Patterns Hiding Across Your Retrospectives

Your team has held retrospective after retrospective. You've captured what worked, what didn't, and what you wanted to change.

But six months later, do you know which issues keep coming back?

This week's workflow uses AI to look across multiple retrospectives at once—helping you spot recurring challenges, improvements that are working, and action items that never quite seem to stick.

📌 Before You Start

The AI tools featured in this workflow are examples, not requirements. Every organization has different approved tools, security policies, and ways of working.

The goal of AI Toolsday isn't to recommend one specific product—it's to share practical AI workflows that you can adapt using the tools available to you.

Focus on the workflow, not the tool.

🛠️ Tools Used

NotebookLM

Bring multiple retrospective artifacts into one place and analyze them together while keeping the analysis grounded in the source material you've provided.

ChatGPT

Take the patterns you've uncovered and turn them into questions, experiments, and potential improvement actions for your team.

⚙️ How It Works

Step 1: Build your retrospective history

Gather notes from your team's last several retrospectives.

Depending on what you capture, that might include:

What went well

What didn't go well

Improvement ideas

Action items

Team health observations

Impediments

Follow-up notes from previous actions

Upload several retrospectives into NotebookLM.

Three months is a good place to start. Six months or more can reveal even more interesting patterns.

Step 2: Look across retrospectives—not at them individually

Now ask NotebookLM questions that require it to look across the entire collection.

Try questions like:

Which challenges appear repeatedly?

Which action items have been raised more than once?

What problems seem to be improving?

What problems appear to be getting worse?

Are there themes around dependencies, workload, quality, communication, or decision-making?

Which issues have generated action items but continue to appear?

What positive themes have become more common over time?

You're no longer asking, "How did the last Sprint go?"

You're asking, "What is our retrospective history telling us about our system?"

Step 3: Pay special attention to repeat offenders

This is where things get interesting.

Suppose your AI analysis discovers some version of "stories weren't ready when the Sprint started" in five of your last eight retrospectives.

Or perhaps cross-team dependencies have appeared repeatedly despite multiple action items.

Ask NotebookLM to identify recurring issues alongside the actions your team previously agreed to take.

This can help you distinguish between a one-time problem and something more systemic.

Step 4: Move the patterns into ChatGPT

Once NotebookLM has identified the major trends, bring that analysis into ChatGPT.

Now shift from pattern recognition to problem solving.

Ask ChatGPT to:

Group related issues

Suggest possible systemic causes

Identify areas that warrant deeper investigation

Generate questions for the team

Suggest small experiments

Identify measures that could help determine whether an experiment is working

Be careful here: AI can suggest possible causes, but it doesn't know why something is happening in your organization.

Treat its ideas as hypotheses—not conclusions.

Step 5: Bring the patterns back to the team

Instead of running your next retrospective as another isolated event, show the team what you've found.

For example:

"Dependency issues have appeared in five of our last six retrospectives. We've tried three different actions, but the theme keeps returning. What's this telling us?"

That opens a very different conversation than simply asking, "What didn't go well this Sprint?"

Your team can decide whether it's time to stop treating the symptom and investigate the system creating it.

💡 Why Agile Pros Care

Retrospectives are designed to drive continuous improvement, but it's easy to focus so heavily on the most recent Sprint that we miss what's happening over time.

AI makes it much easier to step back and look across months of team feedback.

That can reveal recurring obstacles, abandoned improvement actions, positive trends, and systemic issues that deserve more attention.

The goal isn't to let AI diagnose your team. It's to give your team another perspective on its own history.

💬 Prompt to Copy

You are helping an Agile team analyze patterns across several months of retrospectives.
Below is a summary of recurring themes identified from our retrospective history.
Help me explore these patterns without assuming you know their root causes.
For each recurring theme:
Describe the pattern.
Identify previous actions we've tried, if provided.
Suggest possible contributing factors as hypotheses.
Generate questions the team should explore to better understand the issue.
Suggest one or two small experiments the team could consider.
Recommend a simple way to determine whether the experiment is helping.
Also identify:
Issues that appear to be improving
Issues that appear to be getting worse
Recurring issues that may indicate a systemic problem
Positive patterns the team should continue reinforcing
Retrospective Analysis: Paste your NotebookLM findings here.

✅ Try It This Week

Pull together your team's last three retrospectives and ask your approved AI tool:

"Which topics, challenges, or action items appear more than once?"

Pick just one recurring pattern and bring it to your team.

Don't ask AI to solve it. Ask your team:

"Why do we think this keeps coming back?"

💬 Continue the Conversation

Think about your team's retrospectives: Is there one topic that seems to show up again...and again...and again?

What have you learned from trying to address it?

🤝 AI Toolsday Tip

AI is excellent at finding patterns, but a pattern isn't a root cause. Use AI to help you notice what's recurring and generate hypotheses, then let the people closest to the work investigate why it's happening and decide what to change.

Comments (0)

Popular