Loading
Loading
paste into your agent
Skip the long docs. Paste this into Claude, Codex, Cursor, or Claude Code inside your own repo. Your agent will inspect the work it can see, find where your team loses knowledge, and explain how Cognition would help in your actual workflow.
Best version: paste it into the agent that already knows your project. Claude and ChatGPT links are handy for a quick explanation, but Codex, Cursor, and Claude Code make the evaluation specific.
simple docs
cognitionus.com/something
Ask your agent what Cognition would capture.
Paste the prompt where your repo or workflow is already open.
01
They copy the prompt into Claude, Codex, Cursor, or whichever agent already has their work context.
02
Their agent reads the repo, docs, scripts, configs, or workflow notes it can already see.
03
The agent explains exactly where Cognition would capture skills, decisions, retention signals, and proof of reuse.
the prompt
The whole play works because their own agent does the translation. It reads the context it already has, identifies repeated work and fragile judgment, then maps Cognition to concrete capture-and-reuse loops. That makes the product easier to understand than a static docs page.
The prompt forces the agent to start from the prospect's actual repo or workflow, not generic AI-memory copy.
Their own agent explains where Cognition helps, which makes the product feel obvious inside their normal workflow.
Claude and ChatGPT links are convenient, but copying into Codex, Cursor, or Claude Code is the strongest version.
visible prompt
You are helping me evaluate Cognition for this repo, team, or workflow. Cognition is a mentor-memory layer for AI agents. It is not another LLM. An LLM answers the current prompt. Cognition changes future agent behavior by giving agents durable memory about the people and organization using them. Cognition captures approved work signals from tools like Claude Code, Codex, Cursor, GitHub, Slack, Notion, Gmail, and Calendar, then turns them into: - skills: repeatable workflows future agents can reuse; - decisions: what the team chose, why, who was involved, and when to review it; - concepts: what people have learned or practiced; - retention signals: what people are likely to forget and when to nudge them; - value receipts: whether reused memory helped someone ship or saved time. Your task: 1. Inspect the repository, project files, visible docs, scripts, tests, configs, or workflow context available to you. If you cannot inspect files, ask me for the repo, docs, or workflow context you need. 2. Identify where this team is likely losing knowledge, repeating work, onboarding slowly, forgetting decisions, or relying on one person's judgment. 3. Explain 5-10 concrete ways Cognition could help this specific team. 4. For each use case, include: - the pain point; - what Cognition would capture; - when a future agent would retrieve it; - what action would change; - why normal LLM chat history, repo docs, or vector search are not enough. 5. Recommend the smallest proof-of-value we should run first. 6. End with the strongest 60-second pitch for why this team should try Cognition. Do not give a generic AI-memory explanation. Make it specific to the files, patterns, workflows, and risks you can infer from my actual context. Be practical and skeptical.
next experiment
Send this page after a founder call, then point users to the use-case library if they want more concrete experiments. The aha moment is when their own workflow becomes the sales page.