Artificial intelligence enablement is the work of making an organisation able to get value from AI, and in most companies it is being run as a training exercise. Buy licences, book a workshop on prompting, measure attendance, declare the workforce enabled. Six months later nothing measurable has changed and the conclusion drawn is that people need more training. We wrote the plain-language version of what the term should mean in what AI enablement actually means; this post is about why the usual response to it does not work.
The measured picture says the diagnosis is circular. Deloitte's 2026 State of AI in the Enterprise research found the AI skills gap named as the biggest barrier to integration, and found that education, rather than role or workflow redesign, was the number one way companies adjusted their talent strategies in response to AI. The same research found just 34% of organisations are genuinely reimagining the business rather than layering AI onto existing processes. So the most-cited barrier and the most-common response have coexisted for two years without the barrier moving.
That is usually a sign the response is aimed at the wrong thing.
Training teaches the tool. Enablement changes the job.
The distinction is not semantic and it is easiest to see in a concrete case.
Take the person who compiles your Monday operations report. Train them on prompting and they will produce the same report slightly faster, in the same format, with the same three hours of gathering beforehand, because the gathering was never the part a chatbot could help with. Nothing structural has changed. The report still exists in the shape it did when it was designed in 2019, and it is still read by four people, two of whom skim it.
Enablement asks a different set of questions. Why does the gathering take three hours, and is it because the data lives in four systems that do not talk? What decision does this report inform, and does the person making it need the whole document or three numbers? If the compilation becomes near-instant, what should the person do with the recovered time, and has anyone told them?
Only the first of those is a technology question, and it is an AI integration question rather than a training one. The rest are about how the work is organised. This is why the workshop does not move the number: it optimises a step in a process that nobody re-examined.
The four parts of artificial intelligence enablement, in order
Training is the fourth, and it is much cheaper once the other three are done.
1. Pick the work, not the tool. The single decision that determines whether any of this pays back is which task you start with, and it should be a task somebody actually does, frequently, that is shaped like language. We set out the filter in AI workflows: how to find the ones actually worth automating, and the counterintuitive part remains that the flashy customer-facing chatbot is usually the worst first choice, as we argued in the AI projects that quietly pay for themselves.
2. Make the data reachable. Almost every enablement programme that stalls, stalls here. People are trained, willing and licensed, and then discover the assistant cannot see the customer record, the job history or the pricing. Enthusiasm does not survive that for long, and no amount of further training fixes it. This is data foundations work and it is the least glamorous line in the budget.
3. Say what changes, including for the person. Two answers have to exist in writing before anyone is trained. What the recovered time is for, and what happens to the role. BCG's 2026 AI at Work survey found that of frontline employees who use AI regularly, 42% save at least a full working day a week, and 66% of those receive limited or no guidance on what to do with that time. In the same research, an explicit strategy raised measured business impact by 25 percentage points, against five points for better access to tools. The strategy is worth five times the tooling, and it costs an afternoon.
4. Then train, on the actual task. Generic prompting courses age badly and transfer poorly. Training on the specific task, with the real data, alongside the person who does it, is both cheaper and more durable. The output should be a written procedure for that task, not a certificate.
Notice that three of the four are decisions rather than purchases, and that the order matters. Training people to use a tool that cannot reach their data, on a task nobody chose deliberately, with no answer to what happens next, is how enablement budgets get spent without producing anything. That sequence describes most of the programmes we are asked to review. The longer version of what the four parts look like as a programme sits on our AI enablement page.
Who owns it
The most common structural failure is that enablement is assigned to whoever is least equipped to change the work. It lands with IT, who can provision licences but cannot redesign a finance process, or with HR, who can run training but cannot decide what the operations team stops doing.
Enablement needs someone who can change how work is done in a department, which usually means the person who runs that department, with support rather than delegation. In a business of twenty to two hundred people, that person is often the owner, and the honest version of this post is that the work will not be done by anyone else.
There is a related failure worth naming. Deloitte's Tech Trends 2026 research found 38% of organisations piloting AI agents against 11% with anything in production, and attributed the gap in part to organisations automating existing processes rather than redesigning operations. The same instinct that turns enablement into training turns automation into a faster version of a process that was already wrong.
What this is not
It is not a change-management programme, and we would be cautious of anyone selling one at this scale. The activities that move the number are small and specific: choosing one task, connecting one data source, writing down two answers, and training two people properly on the thing they actually do. Deciding what the number is, before any of it starts, is the discipline we set out in measuring AI ROI without lying to yourself.
It is also not an adoption problem in the way that phrase is usually meant. We argued in employee resistance to AI is usually a rational response that when a rollout produces nothing, the likelier explanation is that nobody changed the work rather than that staff refused to. Enablement, done properly, is the name for changing the work. Done improperly, it is the name for a workshop.
If you want to know where your own programme sits, the diagnostic question is short. Ask what specifically people do differently on a Tuesday than they did before the licences were bought. If the answer describes a tool rather than a task, the enablement has not happened yet, whatever the training records say. Working out which task to start with, and what to tell the team about it, is the first conversation in any sensible AI consulting engagement.
FAQ
What is artificial intelligence enablement? It is the work of making an organisation able to get real value from AI: choosing which work AI will do, making the necessary data reachable, deciding and stating what changes for the people involved, and then training on the specific task. It is distinct from AI adoption, which measures whether tools are being used, and from training, which is only the last of the four parts.
Is AI enablement the same as AI training? No, and treating them as the same is the common failure. Training teaches people to use a tool. Enablement changes what the job consists of. Deloitte's 2026 research found education was the most common organisational response to AI while the skills gap remained the most-cited barrier, which suggests the response is not addressing the cause.
Where should a small business start with AI enablement? With one frequently repeated, language-shaped task that somebody genuinely does, and with a written answer to what the time saved is for. Not with a tool selection and not with a workshop. The data connection for that one task is usually the largest piece of actual work.
Who should own AI enablement in a mid-sized company? Whoever has authority to change how work is done in the department concerned, which is usually the person running it and, in a business of twenty to two hundred people, often the owner. IT can provision and HR can train, but neither can decide what a team stops doing.
How do we tell whether enablement is working? Name the behaviour that should change before you start, and check it afterwards. Licences issued and courses completed measure spending, not effect. A usable test is whether you can describe what a specific person now does differently in a specific hour of their week.
Do we need to redesign processes before adopting AI? Not all of them, and not up front. But automating a process nobody has examined produces a faster version of an unexamined process. An hour spent asking why each step exists, before a month is spent automating it, is consistently the highest-return time in these projects.
References
- Deloitte: The State of AI in the Enterprise, 2026 (the AI skills gap named the biggest barrier to integration; education rather than role or workflow redesign as the leading talent response; 34% of organisations genuinely reimagining the business). https://www.deloitte.com/us/en/what-we-do/capabilities/applied-artificial-intelligence/content/state-of-ai-in-the-enterprise.html
- Deloitte Insights: Tech Trends 2026 (38% of organisations piloting agents against 11% in production; automating existing processes rather than redesigning operations as a cause of failure). https://www.deloitte.com/us/en/insights/topics/technology-management/tech-trends.html
- Boston Consulting Group: AI Is Reshaping Jobs Faster Than Companies Are Reshaping Work (press release, 3 June 2026; fourth annual AI at Work survey, 11,749 workers across 14 markets including 503 in South Africa). https://www.bcg.com/press/3june2026-ai-reshaping-jobs-faster-than-companies-reshaping-work
- Boston Consulting Group: AI at Work: Why Strategy Matters More Than Tools (2026; time saved, guidance on recovered time, and explicit strategy against tool access). https://www.bcg.com/publications/2026/ai-at-work-why-strategy-matters-more-than-tools
Written by JP, Sixees Labs. Last reviewed August 2026.