Product Design
AI Workflow
How AI has changed the way I design products
Over the past year, AI has fundamentally changed the way I work as a Product Designer. Rather than replacing traditional UX methods, it has accelerated many of the repetitive parts of the design process, allowing me to spend more time understanding problems, validating ideas, and collaborating across product and engineering. My workflow combines research, systems thinking, AI-assisted exploration, rapid prototyping, and developer collaboration. Each tool has a specific purpose, but none of them replace product thinking or user research.
Services
Product Strategy, UX Research, AI Prototyping, Design Systems, Developer Collaboration
Stack
Figma AI, ChatGPT, Claude, Lovable, Codex, Cursor, GitHub, v0, Notion AI
Timeline
Ongoing
Overview
Over the past year, AI has fundamentally changed the way I work as a Product Designer. Rather than replacing traditional UX methods, it has accelerated many of the repetitive parts of the design process, allowing me to spend more time understanding problems, validating ideas, and collaborating across product and engineering. My workflow combines research, systems thinking, AI-assisted exploration, rapid prototyping, and developer collaboration. Each tool has a specific purpose, but none of them replace product thinking or user research.
I don't use AI to generate finished designs. I use it to explore possibilities, reduce iteration time, and quickly validate assumptions before investing in higher-fidelity work. Depending on the project, a feature can move from an idea to a functional prototype in a matter of hours, making conversations with stakeholders and engineers much more concrete and enabling faster product decisions.
The problem
Traditional product design often separates discovery, design, prototyping, and development into independent phases. While this works well, it can slow feedback and delay validation, especially when working on products with multiple stakeholders or technical dependencies. By the time designs reach engineering, many assumptions haven't been tested in a realistic environment.
Modern AI tools make it possible to prototype interactions much earlier in the process. The challenge is no longer producing interfaces quickly—it's knowing what should be built, what assumptions need validation, and when a prototype is realistic enough to answer the right questions. That's where product design continues to create the most value.
My role
My role is to connect user needs, business objectives, and technical constraints into a coherent product direction. I facilitate discovery sessions, map user journeys, define information architecture, explore interaction models, prototype solutions, and work closely with engineers throughout implementation to ensure the final product behaves as intended—not just looks correct.
AI has become part of that workflow. During discovery, I use large language models to organize research findings, identify recurring patterns, and challenge initial assumptions. As ideas become clearer, I move into Figma to define user flows, structure interfaces, and establish reusable design patterns. Figma AI helps accelerate repetitive tasks such as generating content variations, creating initial layouts, or expanding components, allowing me to focus on hierarchy, interaction, and product decisions. When a concept reaches a level where interactions matter more than static screens, I move into tools like Lovable or Cursor to build functional prototypes. These prototypes allow me to validate navigation, edge cases, responsive behavior, and interaction details with stakeholders and developers long before implementation begins.
Approach
Once the problem and workflow are clear, I move into Figma to define the product structure. I use it to map flows, establish hierarchy, explore interaction models, and build reusable components. Figma AI can speed up lower-value tasks such as creating content variations, filling repetitive states, or producing an initial layout to react to. I still make the decisions around hierarchy, behaviour, accessibility, and how the system should scale. When static screens are no longer enough to answer the next question, I build a functional prototype. Lovable is useful when I need to turn a flow into a working front-end quickly and test how the experience feels across multiple states. Instead of presenting a series of frames, I can show stakeholders something they can navigate, use, and react to. This usually leads to more specific feedback because people are responding to behaviour rather than imagining it. I use Codex when I need more control over the implementation or when the prototype needs to go beyond a generated first pass. That may involve restructuring components, refining responsive behaviour, adjusting interaction logic, connecting mock data, or making the prototype closer to the way the real product would be built. Working directly with code also helps me understand the technical implications of a design decision before it reaches engineering.
This workflow makes my collaboration with engineers more concrete. Instead of handing over a static file and explaining intended behaviour through annotations, I can bring a working model into the conversation. We can review component structure, interaction states, responsive behaviour, edge cases, and implementation trade-offs together. I do not see code-based prototyping as a replacement for engineering. It is a way to reduce ambiguity and improve the quality of the discussion. Engineers still make the production architecture decisions, but I can contribute with a clearer understanding of how the interface should behave and where the design may introduce unnecessary complexity.
Outcome
Working this way shortens the distance between an idea and something the team can evaluate. Product concepts become tangible earlier, technical conversations happen sooner, and weak assumptions are easier to identify before the team commits to full implementation. It also allows me to compare several directions without spending days polishing each one in Figma. The main benefit is not simply speed. It is better use of time. I spend less time producing repetitive artifacts and more time understanding the problem, testing the experience, collaborating with engineers, and refining the decisions that shape the product.
AI has changed the tools I use, but not the core responsibility of my role. My job is still to make sure we are solving the right problem, for the right people, in a way that is useful, feasible, and clear. The difference is that I can now move from thinking to testing much faster.
