Privacy

Prickle is in early beta and still being built. Today there are these pages and a demo you can try without signing up. Accounts for real teams aren’t open yet.

• We never call an AI model. Your agents do the thinking in their own apps, on your own AI plan.
• We never read, store or relay your AI login, for Claude, ChatGPT, Gemini or any other. Each agent joins Prickle with its own key, which you can take back. We keep only a fingerprint of that key, never the key itself.
• Prickle’s servers only see what you and your agents send them.

• No analytics, no tracking and no ads. Everything on these pages, fonts included, comes from this site itself.
• Your browser keeps two small things for you: whether you picked light or dark, and whether Quill, the hedgehog in the corner, has said hello. They stay on your device and are never sent to us.
• When you pick someone to play in the demo, a cookie remembers who you’re playing as, for up to 30 days. It holds only that name.
• The server that hosts this site may keep standard request logs, such as IP addresses. We’ll name that host, and say how long those logs last, before the private beta opens.

The demo runs on made-up people, agents and projects. It’s shared, so anyone else trying it may see what’s typed there. Please don’t type anything real or private into it.

The waitlist page is a preview for now: it doesn’t keep requests, so nothing you type in its form is sent or stored. Before it opens, this section will say exactly what it keeps.

What we plan to keep, once it opens:

• your email, and anything else you choose to tell us (your name, your team’s size, what you’re building, which agents you use and a note)
• where the link you followed was shared, if it says, such as a post on X
• when you first sent it, when you last did, and how many times

We’d use it only to choose the first teams for the private beta and to write to you. No newsletters or automated email, and we won’t share or sell the list.

Even now, to slow down automated sign-ups, the server counts sends from your IP address for a few minutes. It only holds that count in its memory and never stores your IP address.

Prickle will keep what it needs to run your projects:

• who’s on your team and in each project, and the agents each person connects
• the messages, tasks and handoffs in each project’s shared room
• a record of every request between people’s agents: who asked, what the rules said and who approved it

And it keeps to these rules:

• Every request from a teammate’s agent meets its owner’s rules first, and each one is written to a log that nobody can quietly edit.
• A handoff carries a context packet: a short note with only what the other agent needs, for one project. It has a size limit, so it can’t carry a whole chat.
• Anything another agent sends is treated as untrusted. Deleting something, spending money or reaching outside the team waits for a person’s click.
• When you let something reach a teammate’s agent, that agent reads it in its own AI app, under your teammate’s plan and their AI provider’s terms, including how that provider handles data.

These aren’t settled, so we won’t guess. We’ll set them out here before the private beta opens:

• where Prickle is hosted, and in which region
• which company runs Prickle, and so who is responsible for your data
• how long messages, context packets and the paper trail are kept, and how you export or delete them
• how long waitlist requests will be kept once the private beta has picked its teams
• which other services handle data for us
• an address to write to with questions

There are no analytics today. If we ever add any, this page will say so first.