The 5-Step System That Fixes Stalled AI Projects (and Everything Else)

Sometimes, it’s not a tech problem. It’s not a strategy problem. It’s a stuck problem.

That moment where you’re spinning your wheels - with too many options, no clear next step, or worse, the fear that nothing will change even if you try.

I’ve been there. So have my clients.

And I’ve found that almost every breakthrough - in tech, business, or life - comes down to the same five steps.

Let me walk you through the simple system I use with clients, in my business, and even in my personal life when things feel uncertain or overwhelming.

Step 1: What’s the real problem?

Start by clearly stating where you are today. Not just the surface-level frustration, but the underlying pattern.

A CIO I worked with recently thought their issue was poor stakeholder engagement. But once we dug in, it was clear the real problem was that IT was seen as reactive and order-taking - not a strategic partner.

✅ Don’t skip this. If the foundation’s wrong, everything built on top will crumble.

Step 2: What does better look like?

What would “better” actually look like? What’s the dream scenario?

When I started evince Consulting, I wasn’t just chasing billables. I wanted more freedom, more impact, and the chance to help leaders grow. That vision shaped every decision I made.

For your IT team, this might be shifting from ticket-takers to trusted advisors. For your life, it might be finally getting time and energy back.

✅ Be ambitious. A vague goal will get you vague results.

Step 3: What’s been in the way?

Here’s where it gets honest.

What’s stopped you? What have you already tried? Why didn’t it work?

A team I worked with had rolled out a new operating model three times - each time with the same slide deck. It wasn’t the model that failed. It was the execution. They hadn’t changed how people worked, thought, or were supported.

✅ Patterns don’t break themselves. You have to be willing to change.

Step 4: Commit

At some point, you’ve got to decide: are you in or not?

I see this in advisory sessions all the time. That pause before someone says, “Right. Let’s go.” When a CIO realises they can’t keep deferring the hard stuff. Or when a founder stops chasing every new AI tool and focuses on the one that solves a real problem.

✅ 100% in beats 80% effort every time. Make a commitment - even a small one - and stick to it.

Step 5: Take action

Now you move.

Make a plan. Be consistent. Expect failure. Adapt when it comes.

✅ Action builds clarity. Momentum is more important than perfection.

Put It in Practice: AI Strategy and Copilot Adoption

Let’s apply the same system:

1. What’s the real problem?

It’s not a technology problem. It’s not even a training problem. It’s a clarity problem.

No one’s clear on what success looks like. Copilot was launched without an agreed vision, target use cases, or a framework to measure value. It was treated like a tool rollout, not a strategic capability.

Often, the decision to “just get it in” is driven by pressure to show AI progress. But without anchoring that to business outcomes, it creates noise, not traction.

2. What does better look like?

This is where the shift happens.

Better might look like this:

• The legal team uses Copilot to review contracts 30% faster
• The PMO automates status reporting across dozens of projects
• The finance team generates monthly reports with two fewer FTE hours
• Execs get summarised board packs and industry news digests in minutes

These are specific, measurable, and meaningful outcomes. They’re different across functions - and that’s the point. There’s no “one Copilot to rule them all.”

What matters is that your desired outcomes are:

• Business-aligned
• Function-specific
• Tangible enough to test

3. What’s been in the way?

Once we define the vision, the blockers are easy to spot.

For example:

• No enablement plan or context-based training
• No examples of how Copilot applies in their role
• No clarity on data access or privacy
• Fear of saying the wrong thing to the AI - or doing the wrong thing with it
• No time carved out to explore and play

In one case, we discovered that Copilot didn’t work in key tools because the tenant configuration was incomplete. Another client had disabled certain features for security reasons - but hadn’t communicated why.

You can’t overcome blockers you haven’t named.

4. Commit

This is the turn.

You don’t need a 12-month AI transformation program. You need a Copilot Sprint.

Pick 2-3 business functions. Prioritise based on urgency and readiness. Assign a single-threaded leader to each function - not to evangelise AI, but to test it in a real business process and drive feedback loops.

Set success criteria like:

• X minutes saved per task
• Reduction in time-to-insight
• Improvement in response time or quality

And commit to building short, focused use case playbooks with:

• The goal (e.g. summarise meeting notes)
• The prompts that work
• The gotchas to watch for
• Where the AI added value - and where it didn’t

5. Take action

Start the sprint. Share wins. Debrief failures. And build momentum.

In one engagement, a client created a “Prompt Library” across departments - a simple SharePoint site with 10 use cases per team. It became the most visited internal page in weeks. Not because it was slick, but because it was real.

Another ran weekly “Show Us Your Prompt” lunches. 45 minutes of teams sharing what they tried. Nothing fancy. Just real users getting real value, together.

You can scale later. But to scale, you first need signal. Signal comes from action. And action comes from clarity and commitment.

Whether you’re leading transformation, shaping strategy, or launching AI initiatives, this five-step system gives you a way to get unstuck and get moving.

Try it for yourself. Or with your team. And let me know which step is holding you back right now.