Beyond AI Training:
Why HR and IT Must Build Capability Together

An HR head wants employees to become confident with AI. The IT head wants clarity on the platforms, data and controls that will support that ambition. Business leaders want to know what will improve—and when.

Each is asking a necessary question. The challenge begins when the answers are developed separately.

The symptoms are familiar:

1. The cost–benefit question remains unanswered. Investment in tools and training is visible; evidence of better work is harder to find.
2. Learning and licences are out of sync. Some employees finish training without approved access; others have licences but little clarity on how to use them meaningfully.
3. Pilots impress, then stall. A demonstration works, but integration, ownership, controls and support remain unresolved.
4. Every function starts again. Teams solve similar problems independently, creating duplicate effort and inconsistent practices.
5. Confidence exceeds judgement. Employees can generate an answer but may struggle to check its accuracy, recognise bias or decide what information is safe to share.
6. The workshop ends; momentum fades. Everyday pressures return, questions accumulate and promising habits never become established practice.

These concerns sit at the intersection of workforce capability, technology readiness and business execution. Treating them as separate projects leaves the connections unresolved.

The turning point is a shared plan: a clear vision of what AI should achieve, an honest assessment of current capabilities, agreed business priorities and named accountability for turning learning into results. HR and IT must shape that plan together—with business leaders owning the outcomes and responsible AI guiding every stage.

That is the foundation for an organisation-wide capability strategy. The training calendar follows from it.

Agree the destination—and who is responsible for getting there

Before selecting courses, platforms or pilot cohorts, establish the business ambition.

Is the priority faster reporting? Better customer responsiveness? Stronger decision support? Less repetitive work? Greater capacity for innovation?

Each ambition needs a defined outcome, a realistic scope and an accountable owner. “Become AI-ready” is too broad to guide investment or assess progress.

A joint charter should connect:

• Vision and goals: what should change in the way the organisation works, and why?
• Current capabilities: what skills, tools, data, processes and support already exist?
• Business priorities: which opportunities deserve attention first, considering value, feasibility and risk?
• Responsibilities and decisions: who designs learning, enables access, owns the process, approves risk and authorises deployment?
• Evidence of success: which baseline, quality standards and business measures will determine progress?

HR and L&D shape learning, role development and change. IT enables architecture, access, integration and support. InfoSec, privacy and risk teams establish workable safeguards. Business leaders own process outcomes and adoption; finance helps validate the value.

Shared direction must lead to specific accountability. Every priority initiative needs someone responsible for moving it forward and someone authorised to decide whether it should proceed.

Diagnose what is actually missing

A request for “AI training” can conceal several different problems.

One team may lack AI knowledge. Another may struggle with basic data literacy. A third may understand the tool but work with inconsistent inputs, unclear handovers and no agreed process owner.

The response should differ accordingly.

A readiness survey, supported by focus groups, practical scenarios and workflow discussions, should examine six areas:

• AI understanding and judgement: can people recognise suitable uses, limitations, unsupported outputs and situations requiring human intervention?
• Data literacy: can they interpret information, question inconsistent figures, distinguish sources and understand the implications of poor inputs?
• Process understanding: can they explain how work moves, where delays occur, who owns each step and which exceptions matter?
• Access and technology readiness: are approved tools, appropriate licences, permissions and usable data available?
• Responsible AI readiness: do people understand sensitive information, bias, acceptable use, verification and escalation?
• Motivation and working conditions: what interests people, what concerns them and what support or time do they need to apply learning?

Confidence ratings alone are insufficient. Someone may report being comfortable with AI yet fail to recognise an unsuitable data upload or a persuasive answer unsupported by evidence.

Scenario questions reveal those differences. Workflow walkthroughs help establish whether the barrier is a skill, a broken process or a dependency outside the learner’s control.

The output is a readiness and gap map by role and function, connected to a prioritised opportunity list. Some teams need foundational learning. Others need process clarification, access or data improvements before an AI intervention can succeed.

The survey also identifies potential Digital Champions and establishes a baseline for later comparison. Findings should support development and responsible disclosure, rather than penalise employees for acknowledging uncertainty.

Build a capability pyramid with distinct learning journeys

A common strategy requires different learning experiences across the organisation.

Business, IT and InfoSec journeys remain distinct, with shared checkpoints around the same use cases. Business defines the need; technical teams establish how it can work; assurance teams help determine the conditions under which it should operate.

Responsible AI runs through the entire pyramid: leaders set accountability, managers design controls, employees exercise judgement and specialists test and maintain the safeguards.

Progress from awareness to automation and agentic capability

Progression should follow readiness and business need.

• Awareness — understand the possibilities and boundaries. Establish a shared language, approved practices and the judgement to recognise appropriate use.
• Application — improve a real task. Practise framing problems, supplying suitable context, checking outputs and meeting a defined quality standard.
• Automation — redesign a repeatable workflow. Specify triggers, inputs, rules, approvals, exceptions and ownership before connecting the steps.
• Agentic capability — delegate action within explicit limits. Define what an agent may access or change, when approval is required and how failures are detected and addressed.

Not every employee needs to build agents. Not every workflow benefits from AI. Selecting a simpler, more reliable solution is evidence of good judgement.

At each stage, the balance between AI and human intelligence—HI—must be deliberate. “A human will review it” needs to become a named responsibility: who checks what, against which evidence, and with what authority to stop or correct the process?

Use Digital Champions to make adoption local—and learning reusable

Organisation-wide adoption needs people close enough to understand everyday work and connected enough to bring useful learning back into the system.

A hub-and-spoke model provides that connection.

The hub brings together HR/L&D, IT, InfoSec and business sponsors. It maintains the learning architecture, approved resources, technical guidance, governance and measurement.

Digital Champions form the spokes within functions, teams and locations. Their contribution extends beyond answering tool questions:

• Opportunity scouts: identify recurring problems worth solving and help colleagues describe them clearly.
• AI ambassadors: make responsible experimentation approachable and support adoption within the team.
• Connectors: bring business questions to technical or assurance specialists and share relevant practices across functions.
• Knowledge builders: turn tested work into reusable prompts, process maps, SOPs and playbooks.

Selection should consider business understanding, credibility, communication and responsible judgement alongside technical curiosity. Champions need protected time, manager support, expert access and clear escalation routes.

A useful rhythm combines application challenges, expert interactions, peer exchange and management showcases. Each function can nominate its most important problems, then test which solutions could benefit other teams.

The progression is Learn → Apply → Build → Share → Mentor → Multiply.

Local enthusiasm remains connected to common standards. Champions support adoption; access approvals and risk decisions stay with the designated owners.

Make PULSE the connection between classroom learning and workplace culture

The period between formal sessions determines whether learning becomes useful practice.

PULSE provides a continuous learning layer alongside workshops and the LMS: short nuggets, challenges, peer exchange and guided application that keep AI relevant to everyday work.

Its role goes beyond sending reminders.

• Reinforce essential concepts. Revisit evaluation, prompting, data responsibility and human judgement through small, relevant interventions.
• Reveal what teams want to learn next. Interest, questions, attempted challenges and recurring difficulties help identify demand for deeper learning. These signals supplement the survey and conversations; a click alone does not establish a capability gap.
• Introduce new AI possibilities. Curated examples expose teams to emerging tools and approaches, with context on relevance, limitations and approved use.
• Move curiosity into application. A nugget can lead to a functional lab, a workshop or a workplace challenge, with the resulting learning brought back into PULSE.
• Build evidence for ROI stories. Application submissions can capture the original problem, the change attempted, human checkpoints and observed outcomes for validation by process owners.
• Develop a learning culture outside the classroom. Peer contributions, reflection and recognition make responsible experimentation and knowledge sharing part of the working rhythm.

The five learning streams—AI Fundamentals; Prompting & Evaluation; Tools & Workplace Application; Agentic AI & Automation; and Responsible & Ethical AI—keep that experience connected.

Consider a team interested in AI-assisted reporting. A short example reveals demand. A targeted session develops the skill. A workplace challenge tests the approach with approved data. A Champion supports application. The process owner checks quality and outcomes. The validated learning becomes a reusable example.

PULSE helps close the loop between interest, learning, application and evidence. HR gains a clearer view of emerging needs; IT and InfoSec can see recurring friction and questions; leaders gain visibility into how learning is translating into work.

Carry the same problem from discovery through deployment

Design thinking provides the discipline: understand the people and their work, define the problem, explore alternatives, prototype a bounded solution and test it with users.

The stages should produce connected outputs:

A reporting initiative illustrates why continuity matters. The initial request may be for a dashboard. Discovery may reveal that the real constraint is inconsistent source data and repeated manual preparation.

A sound intervention addresses those dependencies, develops the relevant skills, tests the workflow and establishes ownership before wider rollout. Human checks remain explicit at input validation, exception handling and approval.

The same problem stays visible throughout. Learning, technical implementation and change support contribute to a shared outcome.

Make ROI a thread through the journey

The cost–benefit question raised at the start deserves an answer built from evidence.

Agree the baseline before the pilot. Include preparation, checking and rework. Compare similar tasks and retain quality standards.

Then distinguish four levels of progress:

• Capability: can people perform the task and explain the boundaries of responsible use?
• Adoption: are approved practices being used consistently in the intended workflow?
• Operational improvement: what changed in turnaround time, reporting lag, errors, rework, throughput or service quality?
• Realised business value: how was released capacity redeployed, cost avoided, service improved or another business outcome achieved?

Licence usage and training attendance provide context. They do not establish ROI. Time saved creates capacity; its value depends on what happens next.

The investment should include licences, learning, solution development, integration, internal effort, assurance, support and reinforcement. Business owners and finance should validate assumptions and distinguish estimates from observed results.

PULSE application evidence and Champions’ showcases can help build a portfolio of value stories. Each story needs a baseline, a change, a verified outcome and an owner who can explain the result. Engagement signals help guide learning; validated workplace applications provide stronger evidence of progress.

Periodic reviews should support decisions to improve, scale or stop. A pilot that exposes an unsuitable use case can prevent a larger, less useful investment.

Culture is built through everyday decisions

Employees need more than permission to use a tool. They need clarity about expectations, safe ways to practise and confidence that questioning an AI output is valued.

Managers shape that environment through what they reward: speed, certainly, but also accuracy, responsible judgement, collaboration and knowledge sharing. Concerns about changing roles need space for honest discussion and practical development.

A shared strategy gives those behaviours support. The survey establishes where the organisation stands. Differentiated journeys develop relevant capability. Digital Champions connect expertise with local needs. PULSE sustains learning, reveals emerging interests and helps capture application evidence. Pilots and implementation turn that capability into operational change.

AI capability becomes sustainable when vision, people, processes, technology and accountability advance together—and the organisation can demonstrate how its work is improving.