Opinion: Why AI adoption stalls and what Coventry University Group is learning

AI exposes how an organisation really works before it transforms anything. It reveals the silos, weak data and blurred ownership that separate a promising tool from lasting value. Professor Elena Gaura explains how Coventry University Group is using skills, governance and process redesign to turn that exposure into better services, productivity and growth.

The scale of the gap between what AI can do and what organisations gain from it is widely debated. Its causes receive less attention. Limited AI awareness among leaders and weak governance are well-recognised factors. Our experience at Coventry of working with many organisations points to a deeper problem: work is organised around human handovers between siloed teams, using documents and systems that were never designed to exchange data automatically. An AI tool or agent can perform its task perfectly and still add little value when data cannot move, no one owns the end-to-end process or the organisation cannot adapt.

AI creates value when organisations redesign work around people and technology. It may support, drive or automate parts of the process, while people remain accountable for its purpose, exceptions and consequences. The new workflow needs an owner, usable data, proportionate controls and measures linked to gains in time, quality, cost, capacity or income.

Coventry University Group is testing this whole-system approach on itself. We grow champions, offer every employee relevant development and help local teams own change within institutional guardrails. When a use case succeeds, we reuse its process, controls and learning elsewhere. This creates the critical mass for sustained change. The added benefit is taking the lessons of that into our work supporting businesses.

Institutional AI readiness depends on how the organisation operates

Leadership confidence matters. AI is a charged subject, and some uses affect personal data, employment or access to services. Leaders who doubt their own understanding may postpone AI decisions, hand them entirely to technical specialists or impose blanket restrictions. They need enough knowledge to sponsor redesign, challenge proposals and know when to seek specialist assurance.

Silos matter too. A business process may cross several teams while data, permissions and accountability stop at their boundaries. AI depends on well-organised data that systems can access and interpret. In practice, data may be stored in incompatible formats, described differently across teams, trapped in systems that do not connect, or protected by security controls. No AI tool can repair this operating model.

Governance should provide a clear route through risks such as these. At Coventry, we govern centrally and deploy locally: the Group sets policy, controls, approved platforms and escalation routes, while teams own low-risk experimentation and implementation. Uses involving sensitive data, decisions about people, procurement, integration or wider deployment receive formal technical, and ethical risk assurance.

Skills must reach every level

Human-centred adoption begins with the workforce. People engage when AI helps solve a problem they recognise and they have a voice in redesigning the work. General awareness is only a start. Colleagues need opportunities to apply AI in context, question its output and understand its limits.

Across the Group, around 2,900 colleagues have engaged with Essential AI learning since July 2026, and more than 2,000 have attended AI tools training since Christmas 2025. Colleagues from half our organisational units are developing deeper capability through the Level 4 AI and Automation Practitioner apprenticeship. Their work on real organisational needs creates a distributed network of people equipped to lead change.

This combination spreads capability beyond a small central team. It gives colleagues confidence, develops role-specific expertise and protects time for people leading live projects.

Use cases turn interest into evidence

Our AI Adoption Lab supports 13 live use cases across 10 organisational units. Each use case team defines the outcome, baseline and owner before deciding where AI could improve the process. It then tracks impact on quality, staff time and financial effects such as cost, capacity or income. A working AI artefact proves the use case technically; scaling requires measurable organisational value.

Demonstrators often stall because no one owns the wider change: integration, workflow redesign, governance and impact. To bridge that gap, doctoral and postdoctoral researchers work alongside use-case teams, providing technical expertise and sustained, hands-on support from scoping through implementation. This sits within a repeatable support route that combines continuing staff development, approved tools, reusable playbooks and regular drop-in help from trained AI champions.

When use cases expose institutional disconnects in data, systems, procurement or policy, they give leaders a practical agenda for change beyond the remit of individual teams.
Each use case should leave two assets: a local improvement and a route others can reuse. We record the skills, data, approvals, integrations and measures required, so the next team can adapt tested practice rather than start again. This is how local experiments build institutional capability.

A model that can travel

The AI Commons connects the Skills Academy, Adoption Lab, applied research, doctoral training and responsible governance. Together, they provide a route from discovery to use, combining technical expertise with the knowledge of people who own the problem.

The same principle shapes our regional work. We contribute to the West Midlands Combined Authority's AI agenda and chair the West Midlands Local Skills Improvement Plan working group on embedding digital, data and AI. Employers need support throughout the adoption journey, from assessing readiness and selecting use cases to building skills, managing risk and proving return on investment. For smaller businesses, the gap between introductory training and major consultancy support is particularly acute.

Coventry can help fill that gap because we are applying the model and benefitting from it ourselves. We bring research, adoption support and skills provision around a business problem, then share evidence from changes tested in a complex organisation.

No single tool or course will complete an AI transformation. Organisations need to choose worthwhile problems, back the people closest to them, test responsibly and measure the results. Leaders must then remove the barriers to routine use. That is how better work becomes stronger productivity, improved services and value on the bottom line.