The gap
We designed static tools for dynamic human behaviour.
Science fiction spent forty years teaching us what a digital system should feel like: holographic dashboards, gestural interfaces, responsive environments, command centres that anticipate. Systems that understand where you are, know what you are trying to do, and respond fluidly.
What we actually built was filters, forms, captchas, “forgot password”, and chatbots that route you back to the FAQ. Most digital platforms became excellent containers for information and transactions instead of useful partners in helping real people achieving their real goals at the first reason of being.
This is not a failure of ambition or of craft from our team, but simply a structural consequence of the assumptions we designed against. And today, in the age of AI, every one of those assumptions is now wrong.
| We assumed | But |
|---|---|
| Personas as fixed labels | Needs evolve by context, capability, confidence and intent |
| Platforms as repositories | People need guidance across moments, systems and decisions |
| Data as reporting | It could be shared intelligence for continuous adaptation |
| Journeys as printouts | Real behaviour is non-linear, interrupted, and shaped by changing motivation |
The new rules
And then came AI. The Rules of Competition have Changed.
For twenty years, companies competed through better interfaces, better functionality, better digital experiences. Everyone now has access to the same models, and agents can generate interfaces at pace. The capabilities that used to differentiate a product are becoming commodities, quickly.
Five things remain difficult to acquire, because each is earned rather than bought:
- Understanding of your customers — not only their demographics, their trajectory.
- Understanding of your organizational reality — what you can actually deliver, and how fast.
- Understanding of your domain’s complexity — knowing how to make judgement calls in the face of uncertainty.
- Understanding of quality — what “good” means here, and where you are setting the bar higher that your competitors.
- Understanding of outcomes — what progress looks like for the person on the other side.
The model
So what does a platform look like when context becomes the asset?
| Layer | Purpose | Question |
|---|---|---|
| Shared intelligence | Collect context | What do we know? |
| Guidance | Interpret context | What does it mean? |
| Orchestration | Activate context | What should happen next? |
| Experience | Deliver context | How does it show up for the user? |
Layered together, this stack offers a structure for envisioning, planning, and designing digital platforms for the AI era. Every layer depends on the one above it, and each one fails when the layer above is thin. For example, a recommendation engine built on fragmented signals does not throw an error; it just recommends the wrong thing, confidently, for eighteen months.
This is also why most AI initiatives stall at the bottom. Organisations start at the experience layer, because that is the layer with a demo. In reality, the layer that actually creates advantage has no screen, no launch moment, and no obvious owner. It’s invisible, in the backstage.
What follows is each layer in turn: the shift it demands, the questions worth arguing about in your own organisation, and the organisational capabilities without which the layer will not hold.
Layer 01 — Shared intelligence
From fragmented signals to compounding intelligence.
When your AI models become commodities, advantage moves to the layer that makes sense of your context. The winner is not the organization with the most data, but the one with the better ability to interpret what the data it has been accumulating actually means.
Most organizations already have terabytes of events and almost no context, because they instrumented for reporting rather than interpretation. A logged click that does not carry what the person was trying to do is a fact with no meaning. And the context that does exist is fragmented along the org chart rather than around the customer: the call centre knows something the product will never learn.
The shift is from collecting data about behaviour to accumulating understanding of people. This understanding that improves every time your platform is used is what differentiate you and no competitor of yours can buy.
- Every interaction becomes a signal.
- Knowledge accumulates across channels.
- Context is shared across systems.
- Understanding improves with every use.
Organizational
capabilities required
- A single named owner of the customer context model, with authority that crosses channel P&Ls.
- Research operations that maintain a living taxonomy, not an archive of finished studies.
- Consent and data governance designed alongside the experience,
- One shared, written definition of what “progress” means for your customer.
Layer 02 — Guidance
From personalization to radical relevance.
Context only creates value when it can be translated into meaningful action. The next generation of platforms will compete on their ability to understand what context means, and to offer the right guidance at the right moment leveraging these insights.
The distinction is subtle: personalization optimizes for the business’s next best action, guidance optimizes for the customer’s next best step. Those two objectives diverge more often than anyone admits, and a system that can only ever recommend more is a sales engine disguised as an advisor.
This is where the whole challenge is for your organization. It’s not about building a recommendation engine, but building a system that make the right diagnosis based on the context captures — knowing where someone actually is, what is blocking them, and saying “this isn’t for you” when not relevant.
To do so, guidance also needs a model of competence and confidence: two people who want the same outcome need entirely different help depending on what they already know
- Understand intent.
- Diagnose current state.
- Identify barriers to action.
- Optimize for the next best step.
Organizational
capabilities required
- Documented progression models per segment, owned by the business rather than buried in a model.
- A standing evaluation practice for your guidance engine supported by a panel of real cases, re-run every release.
- Subject-matter experts embedded in delivery, because domain judgement is the training data.
- Governance able to approve a system that is allowed to say “I don’t know”.
Layer 03 — Orchestration
From automating journeys to weaving the seams.
The greatest opportunity of AI is not removing humans from the process. It is coordinating people, systems and AI more effectively to build platforms powered by connected intelligence.
Trust towards an organization tend to get lost at the seams, when it people transitions between different steps that feel disconnected — between chat and human, between marketing and service, between the app and the branch. Customer journeys break exactly where budgets do, which is why the organizational chart is always visible in the experience.
AI makes each individual step cheaper, and the temptation is to add more of them. Streamlined and cost-efficient, but poorly integrated. The discipline of winning digital platforms in the AI era is the opposite: fewer, bounded units of work, with explicit checkpoints and a system memory of what was promised to whom.
Otherwise, an orchestration layer that cannot remember a commitment is just routing, not an experience.
- Break work into bounded units.
- Create checkpoints.
- Coordinate humans and machines.
- Design intelligent handoffs.
Organizational
capabilities required
- Documented progression models per segment, owned by the business rather than buried in a model.
- A standing evaluation practice for your guidance engine supported by a panel of real cases, re-run every release.
- Subject-matter experts embedded in delivery, because domain judgement is the training data.
- Governance able to approve a system that is allowed to say “I don’t know”.
Layer 04 — Experience
From designing interfaces to designing grammars.
Users do not experience channels, products or systems, they experience outcomes. As interfaces are increasingly composed at runtime, the role of design shifts from creating screens to defining the principles, rules and guardrails that make every interaction generated by AI feel driven by a single intention.
In that context, the role of design governance shifts from validating artefacts to governing systems. Instead of approving screens, organizations must define the rules, constraints, and standards that ensure every generated experience remains coherent, trusted, and on-brand.
Adaptivity also carries a cost for the users of your digital platform. If the interface keeps changing, people can never learn it. So the grammar has to be tightly connected to your brand to specify what makes you unique and never moves, as carefully as it specifies what may.
- Rules.
- Constraints.
- Adaptability.
- Quality standards.
Organizational
capabilities required
- A design system extended into behaviour, including rules and constraints, not only components.
- A named quality standard for generated experience, with a team sampling against it weekly.
- Brand, legal and accessibility defining guardrails up front rather than approving outputs.
- Leadership comfortable approving a range of outcomes instead of a fixed one.
Conclusion
The new logic: human and machine intelligence.
In the last decade we competed on digital experiences that kept customers engaged. In the next, we will compete on how effectively we transform context into progress for the people using our platforms.
It comes with one condition. Context in the absence of a theory of progress is simply surveillance with better manners. The organizations that will earn the right to know this much about their customers are the ones that can state, plainly and in advance, what they intend to help those customers achieve, and organize their operations to be held to it.
So the real work being the design of future-ready, digital platform is not really an AI programme. It comes back to the foundations. It is the harder work of deciding what your business is for, what role human vs machine hold in the delivery of its value proposition, and then building a platform honest enough to pursue it at scale.
