Beyond the Boardroom: Why AI Success Requires Bottom-Up Innovation

Business Reporter reports that enterprise AI has reached an awkward inflection point: pilots have demonstrated enough potential, but only 1% of business leaders say their organisations have achieved AI maturity, according to McKinsey. The next obstacle is not necessarily legacy IT. It is the less glamorous wreckage underneath: poor data, immature processes and a shortage of useful ideas from the people doing the work. That matters because a strategy drafted in the boardroom cannot see every point of friction on the factory floor, in the finance team or inside customer operations.

The boardroom can set direction. It cannot find every use case.

There is a familiar corporate mirage here. Executives define ambition, approve investment and announce transformation. Then everyone acts surprised when the resulting AI programme produces a collection of disconnected experiments rather than a change in how the organisation actually operates.

The source’s central point is blunt and correct: the most valuable opportunities often sit in ordinary work. Repetitive tasks consume hours. Decisions stall because information lives in disconnected places. Knowledge remains difficult to access. Manual processes survive simply because someone once called them “the way we do things”.

Those closest to the problems are usually best placed to identify where AI could help. Not because employees possess some mystical innovation gene, but because they experience the friction directly. They know which task is tedious, which approval is needlessly slow and which information gap repeatedly forces people to improvise.

That creates a simple test for any AI strategy. Does it give employees a credible route to surface these problems, or does it merely ask them to admire the executive vision from a safe distance?

Governance is leverage, not a brake

Many leaders still treat innovation and control as opposing forces. That is a costly piece of hubris. The argument presented by Andriy Terlyha at Intellias is that good governance enables experimentation to scale rather than strangling it.

Employees are already experimenting with AI, whether their organisations have formally approved that activity or not. The real question, then, is not whether experimentation exists. It is whether it happens inside a secure, trusted environment or through a patchwork of disconnected tools and informal practices that eventually becomes impossible to govern.

The practical prescription is straightforward: approved enterprise platforms, clear guidance on data and security, practical training, and governance that removes uncertainty instead of multiplying it. Strong guardrails can make teams faster because employees no longer have to invent their own rules or guess what is acceptable.

That is a more useful definition of control than simply blocking access. A locked door may reduce visible risk. It also guarantees that the organisation learns nothing about where employees see value.

What companies should examine next

For business leaders, the immediate question is not whether to produce another AI roadmap. It is whether the roadmap connects to operational reality.

Start by asking teams where work repeatedly slows down, where information is hard to retrieve and where manual effort has become normalised. Then check whether those ideas can move into approved experimentation without forcing employees to navigate a maze of unclear policies. The important measure is not the number of pilots announced, but whether successful use cases can become part of standard workflows.

This also changes the investment conversation. Spending on platforms without improving data quality and processes risks building a more expensive version of the same dysfunction. AI does not magically cleanse bad information or repair immature operating models. It can expose those weaknesses with remarkable efficiency.

The dividing line in digital transformation is therefore not ambition. Most companies have plenty of that. It is the connection between executive direction and employee-level discovery — between strategy and the thousands of small frictions that strategy documents never mention.

Top-down AI without bottom-up innovation is not transformation. It is merely well-funded optimism with a governance problem.