A New Method · Vol I

Your AI assistant,
trained on how you
actually work.

Record a quick screen walkthrough. Telos Brief turns it into a ready-to-use AI agent — configured for Cursor, Claude Code, or Codex.

What you need

Your own API keys. Every brief runs on your provider accounts, so those providers bill you for the usage — Telos charges you nothing.

Describe your workflow in chat — an Anthropic key only. About $0.06 a brief.
Record a walkthrough — Anthropic plus OpenAI for transcription. About $0.31 for 15 minutes.

Both need a funded account. Figures are estimates on Claude Sonnet 4.6 and scale with recording length and model — the app shows you the estimate before you start.

Get started →See how it works

No card, no Telos fee · Runs on your own API keys

fig. i

REC · 00:11:42

walkthrough.mov

TRANSCRIPT · 09:24

“So whenever I start a new ticket I open Linear, copy the spec into our PR template, then run the test harness on the relevant package…”

Detected · tool

Linear

Detected · pattern

PR template

Detected · ritual

Test harness

−24dB

→ Telos Brief is listening

Inferred from your recording

Linear ticket flow ·Conventional commits ·Jest + Vitest harness ·PR review checklist ·Storybook conventions ·TypeScript strict mode ·pnpm workspaces ·Tailwind tokens ·Conventional changelog ·Sentry error budgets ·Datadog dashboards ·pre-commit hooks ·Linear ticket flow ·Conventional commits ·Jest + Vitest harness ·PR review checklist ·Storybook conventions ·TypeScript strict mode ·pnpm workspaces ·Tailwind tokens ·Conventional changelog ·Sentry error budgets ·Datadog dashboards ·pre-commit hooks ·

02 — The Method

Three steps from screen recording to a working agent.

No prompt engineering. No fragile workflows. Telos Brief watches what you do, transcribes the reasoning, and produces a configuration your tools already understand.

Record your screen

A 10–15 minute walkthrough of your tools and workflows. Narrate as you go — your usual rituals, conventions, and shortcuts. Uploads are capped at 100 MB, so record at 720p or compress before uploading.

Up to 100 MB · Loom, QuickTime, or in-browser

Upload to Telos Brief

We transcribe the audio, analyse the screen capture, and extract your working patterns — tools, file structures, naming, review rituals.

Processed with your own keys · Live status page

III

Download your agent config

A Cursor, Claude Code, or Codex setup ready to run — with rules, custom commands, context files, and MCP wiring already in place.

.cursorrules · CLAUDE.md · codex.toml

03 — Compatible

Export to the platforms your team already uses.

One recording. Three drop-in configurations. Switch coding tools without rebriefing your AI.

Cursor

.cursorrules · .cursor/

Claude Code

CLAUDE.md · skills/

Codex

codex.toml · agents.md

Sample exports —brief.cursorrulesCLAUDE.mdcodex/agents.md.mcp/telos.json