Assess, deploy, measure: an NHS CIO’s method for scalable AI upskilling

The trusts that are getting real value from AI are the ones that know where their skills gaps are, match the right training to the right people and can prove what that training delivered.

That was the main arc of a recent HSJ webinar, sponsored by and produced with Multiverse, in which The Dudley Group Foundation Trust shared how it is building data and AI capability across a 6,000-strong workforce. The approach breaks down into three phases: assess, deploy, measure.

Assess: build a true picture of your skills gaps

Most trusts, the panel agreed, struggle to formulate an evidence-based answer to a basic question: where are our capability gaps, and how big are they? It’s not unheard of for organisations of several thousand to put out an AI skills survey and base their gap assessment on the less than 50 responses they get back. That’s not a data-driven view of anything.

Ravinder Sahota-Thandi, group chief information officer at The Dudley Group and Sandwell & West Birmingham FTs, took a different route. Her team baselined people in key roles against the Skills for the Information Age framework, mapping where each person sits today and where the trust needs them within a year. Baselining its starting point honestly and mapping it to the trust’s goals enabled Ms Sahota-Thandi’s team and Multiverse to produce a skills development blueprint rooted in reality.

Deploy: get the right training to the right people

Once you know the gaps, the panel warned against falling into a common training deployment trap: trying to fill them by training everyone. The job is not to make everybody a data scientist, which is neither realistic nor a good use of anyone’s time.

Dudley Group’s first pilot took 40 staff from across specialties, followed by a second cohort of around 30. Its goal is to gradually create a network of digital champions who can share their knowledge and shift the culture around them. The trust’s AI academy, as the programme is known, leverages Multiverse’s expertise to match courses to roles – from the basics of Excel through to advanced analysis – while tailored curricula put foundational data literacy before AI. People must understand how what they feed into AI tools influences what they get out.

Because a full apprenticeship is a meaningful time commitment, The Dudley Group is also very precise in the way it selects potential champions: managers sign off on enrolments, learners’ objectives are tied to appraisals and their learning is applied to real projects. All training is funded through the apprenticeship levy that the trust already pays, so the practical ask for managers is time rather than budget.

Measure: prove impact, for staff and patients

Of course, none of this counts without evidence of impact. Ms Sahota-Thandi and Multiverse co-designed surveys that capture baseline skills data against which to benchmark outcomes, with questions tailored to each cohort. Foundational learners are measured on their confidence to perform their role; advanced learners against the specific outcomes they set out to achieve. Without those metrics, even well-intentioned projects can look like a distraction from the day job rather than a priority and lose executive support as a result.

The proof of impact shows up in time saved and in quality of care. Across the NHS, Multiverse reports staff working in Excel save an average of 4.5 hours a week, which at Dudley has meant nurses spending less time reporting and more time with patients. The same skills relieve pressure elsewhere because business intelligence teams, buried in statutory reporting, can hand routine analysis back to people closer to the frontline. Patient outcomes remain the north star, with technology and skills the enablers beneath them.

Start with ‘why AI?’

Overall, the panel offers some time-honoured advice for leaders looking to upskill their workforce to make better, more consistent use of AI: start with the why. Decide the outcome first – shorter waits, time given back, safer use of AI – then fit the training and the funding around it.

The full conversation, including the panel’s advice on scaling capability beyond a single cohort, is available to watch on demand now.