If you’ve been anywhere near an AI project this year, you’ve probably felt the whiplash.
One minute, it’s a board-level priority with ambitious targets. The next, it’s quietly shelved after months of meetings and no tangible value.
You’re not imagining it and industry data suggests that nearly half of AI proofs-of-concept never make it to production. That’s billions in sunk cost and a trail of “lessons learned” decks gathering dust.
The hard truth?
Most AI failures aren’t about the tech. They’re about scope, alignment, and human adoption.
We’re trying to build the skyscraper before we’ve poured the slab.
Start Where You Can Win Fast
When leaders ask me how to “do AI right”, my answer is always the same:
Surface up through your teams the actual painful, measurable problems first, prioritise, and stay focused on executing a fix in 90 days or less.
Call it a productivity pilot, a test run, or even a proof-of-value. The label doesn’t matter. What matters is that it’s small enough to manage and big enough to matter for your team to support, trial, and measure.
This isn’t about dipping a toe in the water. It’s about proving to your exec peers, your board, and your teams that AI can deliver value now, not in some hypothetical future program.
The Productivity Survey
Before you choose your first pilot, run a simple productivity survey across your business. I’ve used this in workshops with leadership teams and it quickly surfaces where value hides.
Ask:
- Where are we burning the most hours?
- Where do bottlenecks frustrate both staff and customers?
- Where is the data clean and accessible?
- Where would a visible win get noticed at the top table?
Score each opportunity. The highest-scoring one becomes your pilot.
Themes of Focus
The right pilot depends on your maturity and access to tools. The starting point for a low-maturity organisation will not be the same as for a team already experimenting with advanced GenAI.
- Copilot pilots – If your organisation is already using Microsoft 365, start with integrated features. These are accessible, low-barrier ways to return time to employees and get them comfortable with AI-driven productivity.
- Sales pilots – Where maturity is higher, agent-based pilots can focus on prospect acquisition, research, planning, outreach, and building strong first meeting engagements with deep prospect insight.
- Operations pilots – Look at automation opportunities in support, finance, or service. These often free up hours in back-office functions that staff feel immediately.
Acknowledge capacity and maturity across your employees. Don’t push pilots outside the bounds of what your people can execute. The goal is not to overwhelm but to build confidence and capability one step at a time.
Why Pilots Win in 2025
- They’re Fast – In 6 to 8 weeks you can go from concept to measurable result, integrating outcomes into existing processes without disruption.
- They’re Affordable – A well-scoped pilot costs a fraction of a large-scale program. More importantly, when coupled with the right maturity training, pilots act as catalysts. They help users grow and then create their own solutions, making adoption self-sustaining.
- They Build Trust – Success creates internal champions who advocate for scale and inspire others.
- They Surface Reality – If data is messy or integration is tricky, you’ll find out early and cheaply.
And perhaps most importantly: pilots give you a story backed not by hope but by data and people’s lived experience. Boards don’t rally behind potential. They rally behind proof.
Make It Agile
One of the biggest traps is treating your pilot as a fixed, final design. The reality is that AI tools are evolving at speed. What works today may not be fit for purpose six months from now.
Keep your ideas agile and transferable. Build internal capability around frameworks and models rather than anchoring everything to a single tool. This way, as AI evolves, your ideas can migrate, adapt, and continue to add value.
Agility beats rigidity in 2025.
Turning the First Win Into a Movement
After your first pilot lands:
- Tell the story in plain business language.
- Share metrics in terms the board cares about such as time saved, costs avoided, revenue unlocked.
- Identify your next two or three candidates and repeat the process.
By the time you’re on your third pilot, you’ve built momentum, confidence, and a practical roadmap. That’s when AI stops being a “project” and starts being part of how you operate.
The Takeaway
The AI winners aren’t the ones with the flashiest roadmaps. They are the ones who can point to a single, visible, high-impact change and say, “We did this. It worked. Here’s what’s next.”
If you want to start that journey, join our AI Productivity Workshop. We’ll run the productivity survey with you, pick your first target, and build a 90-day plan that gets results.
Because the pilot that sells itself isn’t just a project. It’s your AI insurance policy.