Models are maps rather than fixed instructions. They do not make decisions for organisations, but they can show what needs attention. Digital transformation affects strategy, structures and value creation, not only software (Vial, 2019). Their value comes from helping organisations ask the right questions at the right stage. The four models operate at different levels, so they work best as a connected scaffold rather than as competing choices.
The Prosci AI Integration Framework begins with the work people actually perform. It separates tasks into ‘My Work’, which needs human presence, judgement or connection; ‘With Me’ work, where a person collaborates with AI; and ‘For Me’ work, which is routine enough for AI or automation to complete (Prosci, 2026). For example, a leader may keep a sensitive performance conversation as My Work, use AI With Me to organise evidence and draft questions, and automate For Me tasks such as formatting notes or scheduling follow-ups. This makes adoption personal and practical. It also prevents the assumption that every task should be automated. Adoption is stronger when people can see the benefit and fit with their work (Rogers, 2003).
McKinsey’s Rewired framework moves to enterprise scale. It helps organisations check whether the organisation is ready to turn a promising pilot into a reliable way of working. The framework focuses on six connected capabilities: a business-led road map, talent, a scalable operating model, distributed technology, accessible and governed data, and adoption and scaling (Lamarre, Smaje and Zemmel, 2023; McKinsey & Company, 2024). A prototype may work well, but it will not scale if data ownership is unclear, integration is fragile, users lack confidence or organisations have not linked the change to measurable business value.
The Design Council’s Double Diamond provides a clear shape for innovation. Discover and Develop are stages where practitioners and teams widen their view, gather evidence and explore options. Define and Deliver are stages where they narrow the challenge, choose a direction, test it and improve it (Design Council, n.d.). In AI projects, Discover may involve observing a process, interviewing users and reviewing errors. Define converts this evidence into a focused problem statement. Develop explores options such as process redesign, rules-based automation, generative AI or no technology change. Deliver tests the strongest option on a small scale.
The Design Thinking Loop gives practitioners a practical way to learn with users. Stanford d.school describes five modes: Empathise, Define, Ideate, Prototype and Test (Stanford d.school, 2018). Teams can move backwards and forwards rather than treating the modes as a straight line. This matters because an AI solution may be technically impressive but awkward, mistrusted or unsuitable in real work. A low-cost prototype, such as a mock chatbot conversation or a manually simulated AI-supported workflow, allows users to expose weaknesses before significant investment.
The distinction between the models is important. The Double Diamond gives organisations a broad project shape and makes divergence and convergence visible. Design Thinking supplies behaviours and methods for learning with users. Prosci helps allocate work between people and AI, while Rewired checks whether the organisation can repeat and scale the solution. A small team may begin with Design Thinking and Prosci; a transformation portfolio will also need the strategic and operational discipline of Rewired. None should be treated as a compliance certificate or a guarantee of success.
Organisations can combine the models in a simple sequence. First, use Empathise and Discover to understand people, pain points and current performance. Next, use Define to agree the problem, desired outcome and boundaries. Apply the Prosci model to decide which tasks should remain human, be supported by AI or be automated. Use Ideate, Develop, Prototype and Test to compare options and gather evidence. If the pilot is valuable, use Rewired to assess the road map, talent, operating model, technology, data and adoption conditions required for responsible scale.
For example, a complaints team may find that advisers spend too long searching records before responding. Organisations could define the need as faster evidence retrieval without reducing accuracy or empathy. Prosci may classify the final customer judgement as My Work, AI-supported summarisation as With Me and routine case routing as For Me. Design Thinking can test the workflow with advisers, while Rewired highlights the data, integration, skills, governance and change activity needed for wider deployment.
These models do not replace a business case or AI governance. Organisations still need evidence of value, data readiness, security, fairness, human oversight and measurable outcomes. The scaffold should therefore remain iterative: test assumptions, learn from users and performance data, then revisit the model when the task, technology or risk changes.
Action Point
Select one AI opportunity in your organisation and map it through all four models. Define the user problem with the Double Diamond and Design Thinking Loop, classify the tasks as My Work, With Me or For Me, then assess the six Rewired capabilities needed to scale. Record one assumption that must be tested before a business case is approved.