Pretoria, South Africa – September 2026
A business can roll out AI software in a week. Six months later, the harder question is whether people are using it in their actual work.
Research into AI adoption points to a gap between access and everyday use. In a 2025 survey of more than 10,600 employees, Boston Consulting Group found that over three-quarters of leaders and managers used generative AI several times a week, compared with 51% of frontline employees. That gap matters: a tool cannot become part of a workflow if the people doing the work have little reason, guidance, or confidence to use it.
The same survey found that regular use was much higher among employees who had received at least five hours of training and access to in-person coaching. Strong leadership support was also associated with more positive attitudes towards AI. These are survey findings, so they do not prove that training or leadership support alone causes adoption. They do show where organisations can pay closer attention.
An OECD study of firms adopting AI describes a broader set of obstacles, including:
- access to relevant skills,
- usable data and
- advice on implementation.
The challenge is practical. A team may understand AI’s potential but still struggle to identify a suitable process, decide who is accountable for the output, or know what information can be used safely.
What works?
Begin with work people recognise. Show how AI can help with a specific task, let the people who do that task test it, and make support available as they learn. Leaders need enough understanding to set priorities and answer questions. Teams need clear boundaries for security, quality, and human review.
What doesn’t?
Announcing a tool, running a single demo, and treating the rollout as complete. Nor does a high login count tell you whether the tool is useful. If adoption is the goal, it is worth tracking repeat use, listening to why people stop using it, and checking whether the work improves. An incentive can encourage people to try a tool, but the experience must be valuable enough to bring them back.
What we’re doing?
At CohesionX, we are putting those ideas into practice with our customers and within our own business.
Our ongoing Roadshow takes an Agentic AI Executive Briefing to VectorMind customers. We meet leadership teams to discuss what AI agents are, where they might fit, and what opportunities and risks deserve attention. We work through practical use cases alongside governance, security, and responsible implementation. The purpose is to help leaders make informed decisions and take those conversations back into their organisations. We couple the customer visits with regular training for new users.
Internally, the delivery team has an adoption dashboard to see where use is growing and where people may need more support. We have also made time off an incentive when adoption targets are met. Both measures help create momentum. They are starting points, and we still have to listen to our people and ask whether AI is making their work easier or better.
Adoption takes curiosity, guidance, and repetition. The test is whether people find enough value to use the software again tomorrow.