Settlement Eligibility

In 48 hours, I built a working prototype of a self-service eligibility checker for a legal-tech SaaS platform.

The goal:
Transform a static intake form into a guided qualification journey — and demonstrate how AI can accelerate enterprise feature development without sacrificing product thinking.

This project served as a proof of concept for using AI tools (ChatGPT for logic and copy, Codex for code generation) to move from idea to functional prototype in two days.

Legal-tech settlement experiences typically present dense legal context followed by a static intake form, with eligibility revealed only after submission. Users must read, interpret, and determine for themselves whether they qualify before receiving confirmation. This creates friction: complex criteria are self-interpreted, clarity arrives late, and the interaction feels transactional. The issue wasn’t visual design. It was structural. Eligibility was treated as a form, not a decision system designed to guide users toward clarity.

Qualification didn’t need to feel procedural. Instead of asking users to interpret legal language and self-filter, the system could guide them through structured decisions, reveal logic progressively, and deliver outcomes with clarity. Reframing eligibility as a guided journey reduces uncertainty for users while improving submission quality and reducing operational triage. The opportunity wasn’t improving a form. It was turning intake into a scalable decision system.

I treated this as an enterprise feature build under tight constraints: a 48-hour window, no design tooling, and a code-first workflow. ChatGPT helped model eligibility logic and edge cases while Codex accelerated React scaffolding and branching logic. I then refined the structure, hierarchy, and state management to ensure the experience felt clear and trustworthy. AI accelerated execution; I owned the architecture, sequencing, and experience design.

I began with logic, not interface. Before thinking about screens, I needed a clear decision architecture.

Using ChatGPT as a structured reasoning partner, I translated settlement criteria into a formal decision tree, mapped disqualification branches, and surfaced edge cases that could create ambiguity. I sequenced each step intentionally to reduce cognitive load, ensuring that high-impact qualifiers appeared early and dependent logic unfolded progressively. At the same time, I calibrated messaging to maintain legal clarity without creating unnecessary friction or alarm.

By the end of Day 1, the eligibility engine existed as a defined system. The logic was structured, the branches were validated, and the outcome states were clearly articulated. The interface would simply become the expression of that foundation.

With the decision framework established, I shifted into implementation.

Using Codex to accelerate scaffolding, I generated the initial React structure, multi-step navigation, and conditional branching logic. From there, I refined the flow manually, tightening hierarchy, improving pacing between steps, and introducing progress tracking and inline validation to reinforce clarity and momentum. Outcome states were designed to feel reasoned rather than abrupt, each paired with contextual explanation and clear next steps.

AI accelerated the build, but the experience design decisions remained intentional. By the end of Day 2, the prototype was fully interactive — a working eligibility engine that demonstrated both structural rigor and execution speed.

One decision at a time to reduce cognitive overload.

Users always understood where they were in the process.

Eligibility states included explanation and next steps, not just binary results.

Designed as a scalable intake model, not a marketing form.

In 48 hours, the concept moved from structural reframing to a fully interactive React-based eligibility engine. The prototype supported multi-branch logic, conditional sequencing, and contextual outcome states, demonstrating that complex legal criteria could be translated into a guided decision system without losing clarity or rigor.

More importantly, the project validated a broader hypothesis: enterprise features do not require long design cycles to test viability. With structured thinking upfront and AI accelerating execution, it’s possible to prototype quickly while maintaining architectural discipline. The shift from static form to decision system showed how intake can improve user clarity and operational efficiency at the same time.

AI did not design the experience. It compressed build time. The logic, sequencing, and reframing of eligibility as a journey remained human decisions. The real outcome was not just a working prototype, but a repeatable model for rapid, structured experimentation where AI acts as a multiplier and product judgment remains the differentiator.