QR card prompt

Give your agents a brain.

A business card or QR code can send someone here, then into the AI they already use with a Cognition prompt prefilled. The model explains why an agent brain is different from just memory, then turns the current chat into a rough demo of how that brain would answer.

Open in Claude Open in ChatGPT

Best version: prefill the prompt and let the person press send. Auto-submitting into another company's chat product is brittle, and it feels less trustworthy than showing the prompt first.

physical CTA

Card back concept

cognitionus.com/something

Ask your AI why agents need a brain.

Scan, open your AI, and try a scuffed version of Cognition inside the chat.

Scan

The card or QR code opens this page, not a hidden prompt.

Choose

They pick Claude, ChatGPT, or copy the prompt into any model.

Try

Their AI explains the agent brain, then runs a scuffed version inside the chat.

the prompt

Not just memory. Usable judgment.

The whole play works because it lets their own AI explain the product: memory stores context, but a brain learns patterns, checks freshness, and teaches future agents how to act. The chat demo is intentionally scrappy, but the product idea becomes obvious fast.

Visible prompt

The prompt is right here on the page, so the moment feels curious instead of sneaky.

One-tap launch

Claude and ChatGPT prefill links are useful when their web apps accept them. Copy remains the reliable path.

Fast physical CTA

Perfect for business cards, conference badges, table tents, stickers, and founder follow-ups.

visible prompt

Hey. Cognition gives your AI agents a brain.

Explain what that means in plain English.

Use this framing:
- "Memory" usually means storing notes, transcripts, files, or chat history.
- A "brain" means the agent can learn from work, compress what mattered, remember who taught it, notice when knowledge is stale, and reuse judgment later.
- Cognition is not just a memory box. It is a learning loop for agents.
- It watches real work as evidence: prompts, files, decisions, stuck points, tool calls, and outcomes.
- It turns useful patterns into approved skills: when to use them, steps to follow, checks, failure modes, and why the judgment worked.
- Later, another agent can borrow that skill instead of rereading a pile of context.

Make it concrete:
1. Explain how Cognition works, from raw work to reusable agent judgment.
2. Explain how this is different from "just memory" or a normal company knowledge base.
3. Show a small example of what my agent brain could contain: preferences, repeated decisions, workflows, and things I am likely to forget.
4. Then make this chat a scuffed demo of that brain. You do not actually have Cognition installed here, so be honest about the limitation.
5. For the rest of this chat, answer as if you are using a lightweight version of my brain: learn my preferences from what I say, keep a small working model of how I think, adapt your answers, and call out assumptions when you are guessing.
6. Invite me to ask a question so I can see how an agent with a brain would answer differently.

next experiment

Print the URL or QR on a small card. The headline can simply be: "Your agents do not need more memory. They need a brain."

Start a group

Give your agents a brain.

Best version: prefill the prompt and let the person press send. Auto-submitting into another company's chat product is brittle, and it feels less trustworthy than showing the prompt first.

Not just memory. Usable judgment.

Visible prompt

The prompt is right here on the page, so the moment feels curious instead of sneaky.

One-tap launch

Claude and ChatGPT prefill links are useful when their web apps accept them. Copy remains the reliable path.

Fast physical CTA

Perfect for business cards, conference badges, table tents, stickers, and founder follow-ups.

Hey. Cognition gives your AI agents a brain. Explain what that means in plain English. Use this framing: - "Memory" usually means storing notes, transcripts, files, or chat history. - A "brain" means the agent can learn from work, compress what mattered, remember who taught it, notice when knowledge is stale, and reuse judgment later. - Cognition is not just a memory box. It is a learning loop for agents. - It watches real work as evidence: prompts, files, decisions, stuck points, tool calls, and outcomes. - It turns useful patterns into approved skills: when to use them, steps to follow, checks, failure modes, and why the judgment worked. - Later, another agent can borrow that skill instead of rereading a pile of context. Make it concrete: 1. Explain how Cognition works, from raw work to reusable agent judgment. 2. Explain how this is different from "just memory" or a normal company knowledge base. 3. Show a small example of what my agent brain could contain: preferences, repeated decisions, workflows, and things I am likely to forget. 4. Then make this chat a scuffed demo of that brain. You do not actually have Cognition installed here, so be honest about the limitation. 5. For the rest of this chat, answer as if you are using a lightweight version of my brain: learn my preferences from what I say, keep a small working model of how I think, adapt your answers, and call out assumptions when you are guessing. 6. Invite me to ask a question so I can see how an agent with a brain would answer differently.