I want a one-page brief on a topic

A single-page synthesis on any subject, pulled from parallel web research and prior context.

A single-page synthesis on any subject, assembled by parallel research agents and trimmed to exactly what you need to know.

The prompt

Paste this into Claude Code
$ Research the current state of on-device LLM inference and give me a one-page brief I can hand to my CTO.

What the Algorithm does

The Algorithm detects a synthesis task with a hard length constraint, picks Standard effort, and routes to the Research Pack. Research fans out parallel agents across Claude, Gemini, Grok, and Perplexity, then compresses findings into a single page.

What you'll see
$ ♻︎ Entering the DOS ALGORITHM… (v0.0.1) ═════════════🗒️ TASK: One-page brief on on-device LLM inference━━━ 👁️ OBSERVE ━━━ 1/7🔎 REVERSE ENGINEERING:• Wants: current state, one page, CTO-ready• Not wanted: academic dump, fluff, link salad💪🏼 EFFORT LEVEL: Standard | Focused synthesis, tight length ceiling🏹 CAPABILITIES SELECTED:• Research (Standard mode) — parallel web synthesis• ContextSearch — surface any prior DOS work on the topic━━━ ⚡ EXECUTE ━━━ 5/7Research skill spawns 3 parallel agents, each returning a dimension (models, hardware, deployment patterns). Synthesizer trims to one page.━━━ ✅ VERIFY ━━━ 6/7Length under 500 words, every claim sourced, no broken URLs.✓ Done. Brief saved to ~/Downloads/on-device-llm-brief.md

What you get back

A single markdown page: 3-4 tight sections, an opening "so what" paragraph, a bullet list of the current landscape, and a short "what to watch" tail. Every claim is cited with a verified URL. You can paste it into Notion, Slack, or email without editing.

Variations

  • Research the same topic but give me a two-page brief with a competitive matrix
  • Quick research on on-device LLM inference, just the 5 bullets I need
  • Deep research on on-device LLM inference and save to MEMORY
  • Brief me like I'm a product manager, not a CTO

Was this page helpful?