Capability / Design and delivery
AI-enabled workflows with human judgment built in
I help teams move from prompt to production without disconnecting design quality, business approval or the frontend system that must support the product afterwards.
not prompt-only
Prompts begin from established patterns and components.
at every critical step
Designers, stakeholders and engineers can correct the work before it moves on.
across future delivery
The system improves the next project as well as the current one.
AI-enabled does not mean AI-owned
An AI-enabled workflow uses AI where it increases useful speed—researching context, producing first drafts, checking consistency or drafting implementation—while keeping the people with domain expertise responsible for judgment. That distinction matters when a design decision affects customers, operations or a production codebase.
I build workflows that make this division of responsibility explicit. Teams know what the AI can propose, what a designer needs to assess, when stakeholders need to decide and when the work is ready for engineering review.
Prompt-first design starts with a system
Manual design often begins with a blank canvas; prompt-first design begins with a well-framed problem and a dependable set of constraints. The designer prompts the AI with the user need, while the AI retrieves the relevant UX patterns, content guidance and production components.
The result is a proposal that can be evaluated against the organisation’s own standards. A designer checks interaction logic, accessibility, content, edge cases and visual mismatch. They may refine the response by prompting again or intervene directly when the work needs specialist judgment.
Connect Figma and frontend reality
A design library alone is not enough if implementation has drifted away from it. I work with design and development to connect the UX pattern library to the frontend component library, including Figma Code Connect where it helps make the relationship visible. That gives the AI two complementary foundations: the intended experience and the components that can deliver it.
It also gives teams a better handoff. Instead of reinterpreting a static design, the delivery work starts from a shared language that can be inspected, reused and maintained.
Approval is a product capability
Business and stakeholder approval is not a delay to automate away. It is the moment when priorities, constraints and risk become visible. In an AI-enabled workflow, it prevents a technically plausible solution from becoming an organisationally wrong one.
Once approved, AI can support implementation from the final design, grounded in the same libraries used earlier. Engineers and designers then review the output as working product, correcting what the automation cannot responsibly decide for itself.
Why teams ask me to lead this
I bring design leadership and operational thinking to AI adoption. I do not position AI as a shortcut around the team; I design the workflow that lets a team move faster while protecting the standards, accountability and learning that make speed sustainable.
Read the AI-enabled design-to-code case study for the applied model, or explore how I design enterprise AI workflows around broader R&D operations.