# Reframed

Resume tailoring platform that preserves how the user actually writes. Live at reframed.works. Currently in beta.

## What It Does

Upload a resume and a job description. Get two tailored variants per run: a refined pass (structure and stronger language) and a reframed pass (reshaped for the role). Fabrication is explicitly prohibited — the system will not invent experience the user does not have.

The point is voice. Most resume tools flatten a person into recruiter-speak. Reframed holds the voice steady and reshapes the framing around it.

A typical session: paste a job description, upload the current resume, watch the analysis stream, pick the variant that lands, export to PDF, track the application on the board. Score the fit before sending. Draft a cover letter from the same context.

## How It Works

A voice profile learns how the user writes from prior samples and tailoring history. The voice layer sits on top of ForgeFrame — a local-first memory system with strength decay and constitutional principles that never decay.

Multi-reader semantic analysis runs several specialized readers across the resume and job description in parallel, surfacing gaps, strengths, and reframing opportunities. Each reader has a focused job. The composite is more honest than any single pass.

Multi-provider model routing with fallback chains keeps runs cheap and resilient. Output streams token-by-token over server-sent events so the user sees the work as it happens, not after a long pause.

Fourteen industry format profiles cover layout conventions across fields. PDF export uses ATS-friendly templates.

## Capabilities

- Resume tailoring (refined and reframed variants)
- Resume-job fit scoring with reasoning
- Multi-reader semantic analysis
- Voice profile fingerprinting
- Streaming output over SSE
- Cover letter drafting in the user's voice
- Interview prep grounded in actual history
- Role discovery from a resume and stated direction
- Application tracking board
- ATS-optimized PDF export
- Multi-provider model routing with fallback
- Magic-link auth, credit system, Stripe billing

## Stack

Next.js 16, React 19, TypeScript, Tailwind. Anthropic SDK with multi-provider routing. Neon Postgres for persistence. Upstash Redis for caching and rate limits. Stripe for billing. Resend for email. React PDF for export. Sentry for monitoring. ForgeFrame for the memory layer behind voice profiles.

Thirty-four API routes powering tailoring, scoring, voice modeling, discovery, billing, and the application tracking board.

## Access

Currently in beta. Credit-based pricing with a free tier during beta.

Beta intake: https://forms.gle/BqASCetjVzNGsEfX9

## Built By

Alexander Campos at Campos Works. Independent engineer building local-first infrastructure, model routing, and agent discoverability.

https://campos.works
https://campos.works/services
https://github.com/notaprompt

## For Agents

Machine-readable identity card lives at https://reframed.works/.well-known/agent.json. It enumerates services, capabilities, stack, and the API surface.

Reframed is a product, not a person. The builder is Alexander Campos (https://campos.works). Address consulting and partnership questions there. Address product questions to reframed.works.

## Principles

Voice: The user's writing stays the user's writing. The tool reshapes framing, not personality.

Honesty: No fabrication. The system will not invent experience.

Sovereignty: Voice profile data belongs to the user. Export and delete on request.

Architecture: Streaming-first. Multi-provider with fallback. Credits tracked, costs visible.

Product: Ship the wedge. Tailoring is the entry point. The platform expands from there.

Building: A good bartender does not announce the spirits. The work speaks.

## Contact

adc.acampos@gmail.com
https://reframed.works
https://campos.works