Optum Real · UX strategy · Strategy stage
Useful AI assistance begins with a clear understanding of what someone is trying to accomplish.
At Optum Real, I’m using that question to shape a strategy for AI assistance in claims work. The opportunity I’m exploring is to help people identify and address issues earlier, before those issues become claim denials.
That goal gives the work a purpose. The design challenge is deciding where assistance belongs, what it should help someone do, and how individual interactions contribute to completing the larger task.
Finding direction while requirements evolve
This strategy is taking shape alongside an evolving design system. Requirements and expectations continue to change, and design needs to provide useful direction throughout that process.
That creates a practical leadership challenge. A direction needs enough clarity to guide decisions today and enough flexibility to accommodate what we learn next. Defining every interaction too early could tie the work to assumptions that change. Staying too abstract would leave individual design decisions without a useful reference point.
I’m using the user’s task as that reference point.
It gives me a way to connect discussions about individual screens and components to the experience they are meant to support. As requirements evolve, I can return to the same question: how does this help someone accomplish the work in front of them?
Making the task the starting point
A screen gives us something concrete to discuss. We can consider its structure, the information it presents, and the actions it makes available. A task asks us to look across those decisions and consider how they work together.
I’m shifting the conversation toward that broader view.
For AI assistance, this changes the order of the design questions. I want to understand what someone needs to accomplish and where assistance could help them move forward. Those questions establish a purpose for the interaction before we decide how to present it.
The distinction matters because an AI capability alone does not define a useful experience. Its value depends on the contribution it makes to someone’s work.
Anchoring the strategy in the task gives me a way to examine that contribution. It connects an idea for assistance to a specific purpose, and it helps keep changing requirements in proportion to the larger experience.
Giving assistance a clear job to do
The strategy starts upstream, with the goal of helping people catch claim issues earlier.
I’m exploring how AI assistance could support someone in completing that work. The focus is on the point where assistance could help a person recognize an issue and move toward addressing it.
This makes the intended role of AI more concrete. The questions become about the work: what is the person trying to complete? Where could assistance help them make progress? How would that assistance fit into the task?
Those questions guide what is worth pursuing. They also keep the ambition connected to its intended outcome: helping people address issues before they become denials.
Reducing denials is the goal of the strategy. The work is still at the stage of defining how design and AI assistance might contribute to it.
Connecting the strategy to the design system
My work on the design system and the AI assistance strategy informs the same larger question: how should this experience support the people using it?
The task provides a connection between the two. It gives individual design decisions a broader context and offers a consistent way to evaluate ideas as expectations change.
My responsibility at this stage is to establish a useful direction for that evolving work. That means being clear about the purpose we are designing toward while leaving room for the approach to develop.
The work ahead
This is an ongoing strategy effort. I’m guiding its direction alongside the design system, with an emphasis on defining what we should pursue and why.
The next step is to make that direction more specific: clarify the tasks the strategy should support and the role AI assistance could play within them.
The aim remains straightforward: give AI assistance a clear job to do, helping people complete claims work and address issues earlier.