The next challenge is not AI adoption. It is workflow translation.
The hard part is not adopting the tool. It is making messy, tacit workflows legible enough to explain, govern, and improve safely.
A lot of AI discussion still focuses on tools: which model, which assistant, which platform, which licence. But the harder problem is not choosing the technology. It is translating real-world work into something structured enough to be readable, governable, compliant, and, where appropriate, machine-assistable.
That is a bigger shift than people admit.
Why most workflows are still informal
Most organisations still run on workflows that are informal, heavily human-driven, and full of tacit knowledge. People know who to ask, which spreadsheet is “the real one”, where the workaround sits, and when to ignore the written process because the actual process lives in someone’s head. In some places, Excel is not just a tool. It is practically the architecture.
Humans can muddle through that. Machines cannot. Nor, frankly, can good governance.
Why AI ambitions slow down at the workflow layer
That is why many AI ambitions start well enough: some curiosity, some access, some experimentation with chatbots. The slowdown comes when organisations try to connect that enthusiasm to real processes.
At that point, the real issue appears. It is not just whether AI can support the workflow reliably. It is whether the workflow is legible at all. Once you look properly, the mess shows up: inconsistent inputs, unclear decision points, fragile handoffs, undocumented exceptions, local shortcuts, and no shared view of how the work actually functions.
This is the real transition now underway. Not simply from “not using AI” to “using AI”, but from informal human workflow to structured workflow that can stand up to scrutiny, support automation where sensible, and meet the expectations of compliance, assurance, and operational control.
Why this matters even more in regulated environments
In regulated environments, that matters even more. Workflows are not just about efficiency. They are often tied to accountability, auditability, safety, and trust. A workflow that is too messy to explain is usually too messy to automate well and often too messy to govern well too.
That is where a lot of the real work will sit over the next decade: helping organisations move from “this is roughly how we do it” to “this is how it works, this is where the risk sits, and this is what can safely change”.
The next challenge is not AI adoption. It is workflow translation.