Skip to content

Your first chat

A walkthrough of your first conversation: signing up, sending a message, and reading the perspectives.

Sign up

  1. Go to app.autopoetic.ai/signup and pick a plan. Every plan starts with a 7-day free trial, and all plans include access to the Humanist agent. See pricing for the details.
  2. Log in with a passkey or a magic link — no passwords.
  3. You land on your Projects page. Create your first project — a project is the container for conversations, knowledge, teammates, and API keys.

Optionally, open the project's Context section and add a preamble — any history or background the agent should always know ("I'm writing a novel set in the year 1394…") or standing behavior instructions ("Your name is Otto."). Context applies to all conversations in the project. Updating context costs a little usage and takes a minute or two to process.

Send a message

  1. Open Conversations and hit New Conversation. The conversation is created when your first message is sent, and its title is taken from the first message.
  2. Type and send. The response streams token-by-token — no spinner, then a dump. While it works, the status line tells you what's happening: "Gathering perspectives…", "Reading knowledge…", "Composing…".
  3. You can safely leave the page mid-turn. The message is queued, and the conversation resumes when you return.

Watch the perspectives work

By default, each perspective's own short take streams in above the response, prefixed with its emoji. The synthesized answer follows below.

Under the response you'll find:

  • A contribution bar showing each perspective's signed weight for this message — who pushed the answer and how hard.
  • A 3D subjective shape of the message — the perspectives' positions rendered as an interactive shape you can rotate, download as a GLB file, or open in a shareable viewer.
  • A Show thinking toggle for the agent's reasoning trace.

If you want to save usage, the checkbox "Hide perspectives (reduces usage, same quality)" turns off per-message perspective evaluations for you. On any response that lacks them, a Show perspectives button can generate them retroactively.

Keep an eye on usage

The footer shows "X% of plan used". Past your plan's soft limit, messages are still processed but rate-limited (lower priority); at the hard limit, sending returns an error until the weekly period resets or you upgrade. Usage is per customer account, across all projects.

From here

  • Branch a conversation with Fork, rewind with Revert, or make it public with Publish — see chat features.
  • Ask a one-off question without touching a conversation — see Ask mode.
  • Curate what the agent learns — see the knowledge repository.