I came across two pieces this week — a LinkedIn post about a high‑profile AI rollback at one of Australia’s big banks, and Forbes piece, “Why 95% Of AI Pilots Fail.” Both made me pause, not because they prove AI is broken, but because they highlight a predictable human problem: projects that prioritise models over people and outcomes.
Alignment Matters More Than Algorithms
AI amplifies whatever processes it automates. Automating a broken process simply makes mistakes happen faster. MIT’s study found that failed pilots are rarely about model performance — they’re about misalignment. Strategy that lives only in presentations and isn’t shared across sales, marketing, operations, and finance will falter when technology is introduced. The safeguard is a clear, measurable strategy that aligns teams and defines what success looks like before a single line of code or model is deployed.
Trend-Chasing vs. Strategy
Companies often start AI projects for the wrong reasons: to follow a trend, to show they’re “doing AI,” or because a visible use case is easy to pitch. Sales and marketing projects attract roughly 50–70% of AI budgets because they’re simple to explain and measure: chatbots, automated copy, outreach tools. But those easy wins can be superficial. They focus on producing words quickly rather than improving how organisations listen, understand, and meaningfully connect with customers.
Why that matters in practice
Too many AI pilots start with “let’s build a thing” and hope it finds a purpose later. The Forbes analysis names the usual failure modes: technology‑first design, missing stakeholders, vanity metrics and poor workflow fit. The bank story is a reminder that when you ignore people and the systems they work in, you risk trust and reputation — not just a failed experiment.
From day 1, 8 years ago now, our focus has been ‘what are we looking to achieve with our AI tool?’ Coach M wasn’t conceived as another chatbot or an experiment in flashy conversation. It was built on our TLA Methodology: developed to help employees efficiently transform their learning into beneficial, real behavioural change. The aim is not to have longer conversations or more clicks — it’s to achieve learning transfer.
Coach M takes the opposite path. We begin with an action plan where people commit to application goals that are meaningful to them and their context. Coach M talks to learners where people already work — Teams, Slack, email, mobile. Humans stay in the loop: managers, trainers and coaches validate outcomes and handle the nuanced situations AI shouldn’t. That combination reduces the typical failure points: it keeps pilots tightly scoped, encourages real adoption, and focuses measurement on behaviour and performance, not impressions.
Final thought
The point of AI in organisations isn’t novelty or cost-cutting alone; it’s making human capabilities more consistent, frequent and measurable. Coach M is deliberately narrow because narrow works: it solves a clear problem in a low‑risk, human‑centred way.
Photo by Francisco De Legarreta C. on Unsplash
