Every company I talk to wants to discuss AI tools. I make them talk about decisions first. Here's why — and the framework that changes everything.
Here's what I see every week: a CEO or Finance Director reads about AI, feels the pressure, calls a Big 4 consultancy, gets quoted £500K for an "AI Transformation Programme," and quietly shelves the whole idea.
Or worse: they buy an AI platform they don't need, bolt it onto processes that don't work, and six months later conclude "AI doesn't work for our business."
Both outcomes have the same root cause: they started with technology instead of decisions.
Most business leaders sit in one of two camps:
Camp A: "AI will replace everyone." The dystopian view. Leads to paralysis, resistance from staff, and zero progress.
Camp B: "We'll wait until it's proven." The cautious view. Sounds reasonable — until your competitors act first and you're playing catch-up.
There's a third option that nobody talks about: Decision Augmentation.
Build systems where humans and AI make better decisions together than either could alone. AI brings speed and data processing. Humans bring judgment, context, and values. Together, they're better than either.
After 20+ years implementing finance systems and ERP transformations across every industry, I've distilled what works into five sequential steps. Skip one, and you waste money.
Before you think about AI, map the decisions your organisation actually makes. Which ones are slow? Inconsistent across people? Expensive when wrong?
Most organisations have never done this. When they do, they discover 60-70% of their key decisions are repetitive, data-light, and wildly inconsistent depending on who's making them.
That's the goldmine. That's where AI earns its keep.
Not every decision benefits from AI. Score each one on two axes: data availability and judgment requirement.
Technology is the last thing you design. First: who does what, when, and what happens when AI gets it wrong?
For each priority decision, design the handoff: what data does AI see? What does it recommend? How does the human review, override, or approve? What's the fallback if AI is unavailable?
This prevents the two failure modes: over-trust (humans rubber-stamp everything AI says) and under-trust (humans ignore AI entirely, defeating the purpose).
Don't buy a £200K platform before proving the concept. Start with one decision, one team, 30 days. Measure everything: speed, consistency, accuracy, and human confidence in the AI recommendations.
The minimum viable AI might be as simple as a structured prompt workflow, a dashboard surfacing AI-scored recommendations, or an automated analysis digest. Nothing exotic. Nothing expensive.
You'll know within a month whether this works for your business. If it does, you've got the data to justify scaling. If it doesn't, you've invested weeks, not years.
This is where most implementations stall. The pilot worked — now what?
AI that doesn't learn from your people gets worse over time. AI that does gets better. Build the mechanism where human overrides feed back into the system. Train your team not on AI tools, but on when to trust AI and when to override it.
Then expand: take the next decision from your shortlist, apply what you learned, and compress the cycle. Each iteration gets faster.
Every step depends on the one before it. Skip one, and you pay for it:
| Skip this step... | ...and this happens |
|---|---|
| AUDIT | You automate the wrong decisions. Expensive, demoralising, and hard to undo. |
| IDENTIFY | You try to apply AI everywhere. Team burns out, nothing works well. |
| DESIGN | AI makes a bad recommendation, nobody catches it, you lose a client or worse. |
| EXECUTE (too big) | You spend £200K before proving it works. Board loses faith in AI entirely. |
| DEVELOP | Pilot succeeds, then slowly decays. "AI didn't work for us" narrative takes hold. |
The bottleneck isn't AI capability. It's never been AI capability.
It's understanding which of your decisions are costing you money — and building the systems where humans and machines improve together, iteratively, with measurement and feedback.
Start there. Everything else follows.
I offer a free 30-minute Discovery Call to assess where your organisation is and whether the AIDED Framework is right for you. No sales pitch. No pressure. Just an honest conversation about your decisions.
Book a Discovery Call →