Can AI enhance my reflective practice?

There’s been a lot of discussion recently about AI disrupting coaching and psychotherapy, and as a passionate advocate of both, I thought I’d give it a go. But I thought I’d reverse the process. Part of the value that I’ve got from both coaching and therapy is having someone listen without judgement, validate what I’m thinking and feeling, and help me to shape and structure the ideas, experiences and often conflicting layers of meaning that I’m trying to process. So I thought I’d start with what I often do as the client in a coaching or therapy session, and see what AI (in this case CoPilot) does with it.

The start point for reflections was the phrase “action orientated insight”. It provoked quite a strong negative reaction in me, and I was interested in understanding what might be behind that. After all, if you’re going to invest time and energy in action, having it based on insights from data seems like a good thing.

Hypothesis:

AI can enhance the quality of my reflective practice and help me to understand my own emotional reactions in the context of my experience, training and values.

Process:

I waffled at MS Word, using the dictate function to capture what I was saying. I was pretty unfiltered as I would be in a coaching or therapy session.

2. I copied and pasted the transcript of my waffling into CoPilot with the prompt “Summarise”

3. I reviewed the summary.

4. I prompted CoPilot: “Can you ask me coaching questions to challenge my perspective?”

5. I took CoPilot up on the offer to “Turn this into a guided journaling exercise”

6. I dictated my responses to CoPilot’s guided journaling exercise into MS Word, then used the prompt “summarise” to summarise the responses.

Output:

Here’s what CoPilot summarised from my initial ramblings:

Summary:

You’ve been exploring what “action-oriented insights” really mean, especially in the context of learning and development. You’re trying to bridge the gap between raw data (like narrative interviews or platform engagement metrics) and meaningful actions that can be taken based on that data.

Narrative interviews provide rich, free-flowing stories that can touch on deeply personal and philosophical themes. These are hard to quantify but potentially rich in insight.
Engagement metrics (e.g., clicks, time spent, completions) are more concrete but don’t necessarily reflect meaningful learning or behavior change.
You’re grappling with the idea that an “insight” should lead to a clear action—but in practice, this is complicated. Insights are shaped by:
- The questions you ask
- Your own biases
- The context in which the data is collected
You highlight that transformative action often comes from uncovering untold stories or hidden patterns, not just from surface-level data.
Using trust as an example, you show how a survey might indicate a lack of trust, but not explain why—and without that understanding, any action taken may be ineffective or misdirected.

Key Insight:

You’re questioning whether “action-oriented insights” are too simplistic a concept for the complexity of human behavior and organizational culture. Real insight often requires deep interpretation, contextual understanding, and empathy, not just data points.

Here’s what CoPilot gave me for the prompt around coaching questions to challenge my perspective:

Challenging Assumptions

What assumptions are you making about what qualifies as an “insight”?
How do you define “action” in your context — and does it always need to be immediate or measurable?
What if the most valuable insights aren’t immediately actionable — how might that shift your approach?

🧠 Exploring Perspectives

Whose perspective is missing from your current interpretation of the data?
If someone from a completely different field (e.g., anthropology, design, or behavioral economics) looked at your data, what might they see that you don’t?
How might your own preferences for narrative or metrics be shaping what you consider “valid” insight?

🔄 Reframing the Problem

What would it look like if you treated “action” not as a solution, but as a hypothesis to test?
How might you design your data collection to invite unexpected insights, rather than confirm what you already suspect?
What if the goal wasn’t to act on every insight, but to sit with some of them longer — what might emerge from that?

🌱 Future-Focused

What kind of insights would you want your future self (or team) to be able to act on — and what would you need to start doing now to make that possible?
If you could redesign your insight-to-action process from scratch, what would you keep, discard, or invent?

Here’s what CoPilot gave me in terms of a guided journalling activity:

Guided Journaling: Exploring Action-Oriented Insights

Instructions:

Set aside 20–30 minutes of quiet, uninterrupted time.
Write freely — don’t worry about grammar or structure.
Be honest and curious. Let your thoughts wander and surprise you.

🧠 Part 1: Challenging Assumptions

1. What do I currently believe an “insight” is?
Where did that belief come from? How has it served me — and how might it limit me?

2. Do I believe every insight must lead to action?
Why or why not? What happens when I encounter insights that don’t suggest a clear next step?

👁️ Part 2: Exploring Perspectives

3. Whose voices or perspectives are missing from the data I usually work with?
What might they see that I don’t?

4. How might my own preferences (for narrative, metrics, etc.) shape what I notice — and what I ignore?
What would it take to notice differently?

🔄 Part 3: Reframing the Problem

5. What if “action” was not a solution, but a hypothesis?
How would that change the way I approach insights?

6. What stories or patterns am I not seeing because I’m not asking the right questions — or because I’m not listening in the right way?
What would it look like to go looking for the “untold” stories?

🌱 Part 4: Future-Focused

7. What kind of insights do I want to be able to act on in the future?
What kind of data, relationships, or practices would I need to cultivate now to make that possible?

8. If I could redesign my insight-to-action process from scratch, what would I keep, discard, or invent?
What’s one small experiment I could try this month to move in that direction?

Here’s CoPilot’s summary of my responses to the guided journaling exercise:

What Is an Insight?

An insight is a perspective shift — something that helps you see with fresh eyes.
It goes beyond the obvious or surface-level data, offering rich interpretation and meaning.
Your understanding of insight is shaped by your background in social science, coaching, and education, where insight often leads to understanding, not necessarily immediate action.
You value insights that illuminate hidden stories, especially those that are excluded by majority-driven summaries or quantitative data alone.

🔄 Tension with Action-Oriented Insights

You don’t believe every insight must lead to action, especially not in a linear or transactional way.
However, you do believe insights should deliver value — whether through deeper understanding, new perspectives, or eventual change.
In business consulting, clients often expect measurable impact, so insights must often translate into clear direction and outcomes.

🎯 Two Types of Action-Oriented Insights

You’ve developed a helpful distinction:

Strategic Insights
- Set the trajectory, define goals, and clarify what needs to change.
- Example: “We need to increase footfall to drive revenue.”
Tactical Insights
- Define the specific steps to achieve the goal.
- Example: “Customers are deterred by long queues — streamline checkout.”

This distinction helped you reframe your initial resistance: your work does produce action-oriented insights — just not always in the narrow, outcome-driven way the term is sometimes used.

🗣️ Whose Voices Are Missing?

You focus on amplifying minority or underrepresented voices — often those with less power in organizations.
You’re cautious of how majority narratives can obscure complexity and silence important stories.
You recognize a need to balance deep, narrative insights with broader data that can guide scalable decisions.

🧪 Action as Hypothesis

You strongly resonate with the idea that action is a hypothesis, not a fixed solution.
You already work in an experimental, iterative way, testing assumptions and learning from outcomes.

🌱 Your Integrated Perspective

You’re moving toward a both/and model: combining deep, storied insights with strategic direction and measurable goals.
Insight, for you, is about fair representation, meaning-making, and informed action — not just data-driven decisions.

Was it useful / what did I learn?

The summary was accurate in terms what I said, and definitely reflects my own beliefs around what “good data” looks like. However it was the next step, asking for coaching questions, that things got interesting. The questions suggested to help me challenge my perspective were really useful and helped the reflection go a level deeper. The guided journaling and the summary led to a meaningful breakthrough for me. I knew that there was value in insights from large sets of data, but I was struggling to connect it to the granular actions that I am often involved in helping organisations to identify, and which I think of as insights. The skepticism about high level insights usefulness to me in my practice was blinding me to the opportunity that exists when you consider insights to have different meanings at a strategic and tactical level. I think I was aware of this, but wasn’t able to articulate it crisply – the articulation is the breakthrough.

What was interesting, in the process of using AI to summarise, is that CoPilot got a bit carried away and added in things to the summary of the journaling exercise that I did not actually say. They were reasonable additions, but I didn’t say them. In the section on Two Types of Actionable Insights, CoPilot summarised:

“Strategic Insights

Set the trajectory, define goals, and clarify what needs to change.

Example: “We need to increase footfall to drive revenue.”

Tactical Insights

Define the specific steps to achieve the goal.

Example: “Customers are deterred by long queues — streamline checkout.””

I didn’t say the text in pink, but it is a reasonable example.

The process of having AI support my reflective practice around this area was useful, and it’s definitely a process I will repeat when I need to challenge myself around why I am reacting to something, and where I need to put some distance between myself and my thoughts so that I can look at them with fresh eyes and get a new perspective.

It was certainly faster and cheaper than coaching or therapy, but did it do the same thing. I’d say no, but not with strong certainty. It was a little bit like taking something to supervision, and was really effective at some levels. What it didn’t get into, which supervision and some types of coaching or therapy might, is why this idea of “action orientated insights” triggered something in me. A coach or therapist might have got into who said it, what that person represented for me, and what experiences were triggered, which may not have belonged to the phrase itself. These questions might have been based on knowing me and my history.

However, replacing coaching and therapy was not the aim of the experiment. Did it “enhance the quality of my reflective practice and help me to understand my own emotional reactions in the context of my experience, training and values”. Absolutely, impressively, and they led to some interesting insights.

How might I build on this experience? What I take from the experiment is that AI has the capability to support good quality reflection on the basis of simple prompts and to suggest effective prompts. Repeating this process could be a useful addition to a program that invites people to explore an aspect of organisational life that they are keen to change, and to reflect on their perspectives about.