Reclaim

RECLAIM uses AI to turn consumer complaints into clear action plans by organizing evidence, finding relevant rules, and guiding users from problem to resolution in one place.

Comment

Inspiration

Consumer disputes can be exhausting. A cancelled flight, rejected insurance claim, or billing problem can leave people searching through documents, policies, and disconnected websites just to figure out what to do next.

We built RECLAIM to make that process simpler: turn a confusing complaint into an organized, evidence-backed path toward resolution.

What it does

RECLAIM is an AI-powered consumer dispute resolution platform that helps users move from problem to action in one place.

Users can describe a dispute, organize supporting evidence, analyze relevant documents, explore applicable regulations, and identify practical next steps. Its agent-based workflow is designed to break complex cases into manageable stages instead of treating every problem as a simple chatbot conversation.

The goal is straightforward: help people understand their options and act with greater confidence.

How we built it

RECLAIM combines a Next.js and TypeScript frontend with a FastAPI backend. Its architecture uses LangGraph to structure AI workflows and Pydantic to validate data.

The system is designed around case intake, document analysis, regulatory research, evidence organization, and decision support. These components work together to turn unstructured complaints into structured cases and actionable recommendations.

We focused on a modular architecture so additional data sources, service integrations, and resolution workflows can be incorporated over time.

Challenges we ran into

One major challenge was turning vague, unstructured complaints into useful and consistent case information. Important details can be buried in lengthy documents, incomplete descriptions, or conflicting evidence.

Another challenge was making AI-generated recommendations useful without presenting uncertain information as fact. Connecting evidence, relevant rules, and suggested actions requires careful validation and clear separation between verified information and AI-generated conclusions.

We also had to balance a simple user experience with a backend architecture capable of supporting more complex workflows.

Accomplishments that we're proud of

  • A resolution-first approach: RECLAIM focuses on helping users determine what to do next, not merely generating answers.
  • Structured AI workflows: Complex disputes can be organized into distinct stages for analysis and decision support.
  • Evidence-centered design: Documents and relevant information are treated as essential inputs to the resolution process.
  • A modular foundation: The architecture is designed to support additional dispute categories and integrations.
  • A practical problem to solve: We are applying AI to a real-world frustration that affects consumers across multiple industries.

What we learned

Building RECLAIM reinforced that useful AI is about more than producing convincing text. It requires structured inputs, reliable evidence, clear reasoning, validation, and practical next steps.

We also learned the importance of distinguishing a promising recommendation from a verified outcome. Trust depends on transparency, and automation should not hide uncertainty from the user.

What's next for Reclaim

Our next steps are to strengthen evidence verification, expand regulatory coverage, improve multilingual document processing, and integrate more real-world service providers.

We also aim to improve recommendation evaluation, add clearer progress tracking for each case, and enable users to review and approve consequential actions before execution.

Our long-term vision is to make RECLAIM a connected consumer resolution platform that helps people navigate disputes with less confusion, fewer repetitive steps, and greater control.

RECLAIM — From problem to proof to resolution.

Built With

Share this project:

Updates

Submission history