Independent product · 2026
Near Fix.
A quote-based local-services marketplace pilot that helps customers discover eligible nearby providers, compare offers, and move from request to completed job.
Marketplace workflow
I built the current pilot across customer, provider, and admin roles. Customers save a location, discover eligible providers by service area and occupation, create direct or broadcast requests with timing and private images, compare quotes, accept one offer, follow job status, and rate completed work. Providers complete onboarding and KYC, publish moderated services, quote or decline requests, and manage assigned jobs; admins review applications, catalog data, public content, ratings, operations, and audit history.
Request Copilot
The Request Copilot turns a customer’s plain-language problem into a reviewable request draft. LangChain and Azure AI Foundry can recommend only from a numbered, server-provided catalogue and available service context. Pydantic validates the structured response, the server resolves real identifiers, unsupported requests are handled explicitly, and the customer remains responsible for approving and submitting the draft.
Reliability decisions
Idempotency keys prevent duplicate request creation, PostgreSQL row locks serialize quote acceptance, and explicit event-backed status transitions keep the request-to-job lifecycle predictable. Eligibility considers service areas, occupations, availability, KYC, and review state. Request images and KYC documents use private upload flows, public profiles and ratings are moderated, and contact details unlock only after a quote is accepted.
Product boundaries
Near Fix is presented as a current marketplace pilot, with final live QA and production hardening still pending. Payments, payouts, chat, calling, maps, and live tracking intentionally remain outside the current scope so the core discovery, quote, assignment, and trust workflows stay focused.