ReclaimDesk AI

An AI warranty recovery agent that helps commercial kitchen service companies recover every warranty dollar they've already earned automating claims, follow-ups, disputes, and reimbursements.

Comment

ReclaimDesk — AI Warranty Revenue Recovery

Inspiration

Every day, commercial kitchen equipment service companies repair ovens, refrigerators, ice machines, and dishwashers under manufacturer warranties. The repair is completed, but getting reimbursed is a completely different battle.

Warranty claims require technician notes, invoices, photos, serial numbers, labor calculations, manufacturer-specific forms, emails, and constant follow-ups. Many companies lose thousands of dollars every month simply because claims are submitted late, rejected for missing information, or never filed at all.

We realized this isn't a paperwork problem — it's a revenue recovery problem. That insight inspired ReclaimDesk.

Instead of building another warranty management dashboard, we built an AI employee whose only job is making sure businesses collect money they've already earned.

What it does

ReclaimDesk is an AI-native warranty recovery platform for commercial kitchen equipment service companies.

It automatically:

  • Reads technician reports, invoices, PDFs, emails, images, and voice notes.
  • Extracts equipment model numbers, serial numbers, labor hours, and replaced parts.
  • Checks warranty eligibility using manufacturer-specific rules.
  • Identifies missing evidence before submission.
  • Builds manufacturer-ready warranty claim packets.
  • Tracks claims through submission, follow-ups, disputes, and reimbursement.
  • Shows exactly how many dollars are ready to recover today.

The product's north-star metric isn't tickets processed or hours saved — it's Warranty Dollars Recovered.

How we built it

We designed ReclaimDesk as an agentic AI workflow instead of a chatbot.

Frontend

  • Next.js 16 + React 19
  • TypeScript
  • Tailwind CSS v4
  • shadcn/ui
  • Framer Motion
  • Recharts
  • TanStack Table

Backend

  • Supabase
  • PostgreSQL + Prisma
  • Server Actions + Route Handlers

AI Layer

  • Freshworks Agent Studio architecture.
  • OpenAI Responses API wrapper.
  • Multi-agent orchestration layer.
  • ElevenLabs voice architecture for technician voice intake.
  • Dodo Payments integration layer for subscriptions.

AI Agents

The workflow is split into specialized agents:

  1. Intake Agent
  2. Extraction Agent
  3. Eligibility Agent
  4. Evidence Agent
  5. Claim Builder Agent
  6. Submission Agent
  7. Recovery Agent
  8. Dispute Agent

Each agent performs one responsibility and writes its progress into a shared timeline, making the entire recovery process transparent.

Challenges we ran into

The biggest challenge wasn't extracting information with AI—it was modeling a real warranty workflow.

Different manufacturers require different combinations of photos, invoices, serial numbers, labor caps, and submission rules. We had to create a flexible claim pipeline that could support multiple OEM policies instead of a single generic workflow.

Another challenge was balancing automation with human approval. Warranty claims involve real money, so users need confidence and visibility before AI submits anything. We redesigned the experience around evidence, confidence scores, and approval checkpoints.

Finally, we wanted the app to feel like a modern startup product—not enterprise software from 2012. We spent significant effort designing a clean, fast, money-first interface inspired by Stripe, Linear, and Ramp.

Accomplishments that we're proud of

  • Built a complete AI-native workflow instead of a single chatbot.
  • Created a product focused on recovering revenue rather than generating documents.
  • Designed a premium SaaS experience with dashboards, analytics, recovery pipelines, inboxes, and AI timelines.
  • Built reusable architecture for future Freshworks MCP integrations.
  • Turned an invisible operational pain into a measurable business KPI.

What we learned

This project taught us that the best AI products don't replace users—they replace repetitive departments.

We also learned the importance of first-principles thinking. Rather than asking "Where can we use AI?", we asked "Why does this business lose money after completing a repair?" That question led us to warranty recovery instead of warranty management.

We also learned how to design AI systems that are explainable. Every decision needs evidence, confidence, and a clear audit trail if businesses are going to trust automation with financial workflows.

What's next for ReclaimDesk

This hackathon is the starting point—not the finish line.

Our roadmap includes:

  • Live Freshworks Agent Studio and MCP integrations.
  • Real manufacturer warranty policy engine for dozens of OEMs.
  • Technician voice intake powered by ElevenLabs.
  • Automated email and portal submission adapters.
  • Warranty Leakage Audit for historical claims.
  • Accounting reconciliation and payment tracking.
  • Usage-based SaaS billing with Dodo Payments.

Our goal is to become the AI warranty department for field-service businesses and eventually expand beyond commercial kitchens into HVAC, refrigeration, and industrial equipment service.

Built With

  • agents
  • css
  • document
  • dodo
  • elevenlabs
  • freshworks
  • intelligence
  • mcp
  • next.js
  • ocr
  • openai
  • payments
  • postgresql
  • prisma
  • radix
  • react
  • recharts
  • shadcn/ui
  • supabase
  • tailwind
  • tanstack
  • typescript
  • ui
  • vercel
Share this project:

Updates