Your AI strategy is sound. What proportion of it has become something your people can use?
Most enterprise AI strategies are well reasoned. Comparatively few survive contact with their people. The distance between the two is a design problem & it is the one I solve.
The premise
Every transformation starts the same way. A current state, a future state and an intent to move between them over time.
Traditional
Modern
My Approach
A defined position
Strategy decides what to pursue. Engineering decides how to build it.
Between them sits a narrower discipline: deciding what should exist at all, what it has to do for the person using it, and what standard it has to meet before they will trust it. That work is usually assumed rather than assigned, and the gap only shows up after delivery, when adoption doesn't follow.
A tool people don't trust doesn't sit unused. It gets worked around, usually in a way the organisation can't see or govern. The reputational exposure isn't in the tool you built; it's in whatever your people reached for instead. When the people who have to use a tool have helped shape it, they use it as intended and inside the limits it was given, because the limits make sense to them.
That's the position. Set the standard before the first line of code, build it with the people who will live with it, and the same destination costs far less distance — the work isn't relitigated after delivery, and people reach the answer sooner.
Services
Engagements scoped to the decision you are facing
Facilitated discovery with the people who do the work and the people accountable for it. We map the workflow, locate where expert judgement is being consumed by process and test each candidate against feasibility, risk and the standard the organisation is known for. You finish with a prioritised, defensible set of opportunities, and a clear account of what was ruled out and why.
A defined picture of what AI-augmented work looks like in your organisation: the workflows that change, the role the human keeps, and the experience the tool has to deliver. Articulated through narrative and visualisation so that executives, engineers and end users are working from the same understanding rather than three compatible-sounding ones.
Working prototypes built rapidly and tested with real users. The outcome is evidence: a validated experience direction, a clear build sequence and an early, honest read on whether the concept deserves to proceed. Deciding not to build something is frequently the most valuable result an engagement produces.
Translating responsible AI intent into the interaction itself. How a tool communicates uncertainty, handles failure, shows its working, and keeps a person meaningfully in control of the decisions they answer for. Governance defines the obligation; this is the design work that makes it observable to the person using the system, and demonstrable to the people assuring it.
Reusable interaction patterns, component foundations and experience standards that give a growing set of AI tools a coherent and credible feel. This is the compounding asset: it prevents a portfolio fragmenting into unrelated products, encodes responsible-design decisions so they are inherited rather than relitigated, and makes each new tool faster to build than the last.
Fractional or program-length leadership of the experience thread across multidisciplinary teams: research, interaction design, prototyping, delivery. Suited to organisations building AI capability faster than they can build the design function to support it. The intent is always to leave the practice behind, not the dependency.
A week of discovery, a prototyping sprint, or an embedded design lead across a program. If you are not sure which, the conversation is free.
Formed, not installed
Capability is formed. It cannot be installed.
Anything an organisation genuinely absorbs — a skill, a standard, a way of working — takes longer to establish than it does to describe, and the time is not waste. It is the mechanism. Deliberate work at the beginning is what makes the result durable at the end.
Enterprise AI has inherited the opposite instinct. The technology arrives quickly, so the response is expected to be quick too. Tools are deployed before anyone has decided what good looks like, and the shortfall surfaces later as low adoption, rework, and a quiet loss of confidence that is far more expensive to recover than it would have been to avoid.
My practice runs on the older discipline. Understand the work before designing for it. Make the idea tangible early, while changing it is still cheap. Decide the standard deliberately, and encode it so it holds as the portfolio grows. I have spent two decades watching digital design mature from surface to system to infrastructure. What is happening now, AI reshaping how knowledge work is actually performed, is the most consequential version of that shift I have seen. It deserves to be approached with more care than speed, not less.
I turn what is unclear into what's possible & connect people to it.
Principles
Four things I hold to
Design augments capability. It never transfers accountability.
An AI tool can extend what a professional is able to do. It cannot assume responsibility for the judgement they are accountable for. Designing that line clearly, where the system assists, where the human decides, and how the person can tell the difference, is the foundation of responsible AI in practice rather than a compliance layer added at review.
Start with the work, and the people doing it.
The strongest opportunities come from understanding a workflow well enough to see where expert judgement is being spent on things that don't require it. The weakest start with a capability and look for somewhere to apply it. The people doing that work are not a research input — they are the reason a tool will or won't be used.
Make it tangible while it is still cheap to change.
A document creates agreement in the room and ambiguity outside it. A working prototype creates a shared object, something executives can react to, users can break, and engineers can build from. Getting there fast is not a shortcut; it is how the expensive decisions get made on evidence.
Never trade quality for speed.
Organisations built on expertise are known for the standard of their work. Augmentation that produces more output at a lower standard doesn't save time, it borrows it, at interest. The goal is better work in less time, with the reputation intact.
The outcome
Regardless of scope, the same three things
A better experience of the work
In an environment producing more data and insight than any individual can reasonably hold.
Time returned
To the person doing the work and to the organisation around them.
Trust is the new standard
To deliver work that is still recognisably yours. Tools your people use, from an organisation who earned it.
In practice
A small glimpse towards the future
Robert Rulli
Principal AI Design Lead
I have spent the last several years designing AI products inside a large professional services environment, working at the point where emerging technology, enterprise governance and real human workflows collide.
That experience shaped a specific conviction: the hardest part of enterprise AI isn't capability, it's translation. Translating strategy into something buildable. Translating a model's behaviour into an interface someone trusts. Translating a pilot into a practice the organisation actually keeps.
My background is design, twenty years of it, from industrial and brand design through UX, immersive technology, and now AI product experience. That range is useful in the era of Human Centred Design. Designing for AI means designing for ambiguity and that is a generalist's problem as much as a specialist's.
Professional and knowledge-intensive services
Designing AI tools for organisations whose product is expertise, where augmentation has to respect professional judgement rather than replace it.
Enterprise AI portfolios
Experience strategy and design foundations for organisations moving from scattered pilots to a coherent, scalable set of AI tools.
Future of knowledge work
Vision and concept work for leaders trying to understand what AI-augmented work looks like in their context, before committing to a build.
B. Design (Industrial Design), First Class Honours, UTS. Postgraduate study in user experience and immersive design. Foundations of Humane Technology, Centre for Humane Technology.
Based in Newcastle NSW, working with clients across Australia and internationally.
Let's talk about what you're trying to build.
Whether you have a defined brief or an ambition you are still shaping, a short conversation is usually enough to work out whether I am the right fit.
Enquire