Documentation Index
Fetch the complete documentation index at: https://docs.thor.ai/llms.txt
Use this file to discover all available pages before exploring further.
AI shouldn’t wait for a prompt
Think about how you actually use AI today. You open ChatGPT. You type a question. You wait. You read. You close the tab. You open Cursor. You highlight some code. You write a prompt. You wait. You apply the diff. You spend an afternoon wiring up an n8n flow or a CRON job. You hope you anticipated the conditions that matter. You forget about it. Three weeks later it breaks silently and nobody notices. Every interaction starts with you deciding something is worth asking about. That’s a problem — because the things that matter most are usually the ones you don’t yet know to ask about. The blocker that just appeared in a thread you’re not in. The teammate who quietly missed a commitment. The decision that got reversed in a meeting you skipped. The customer issue three different people mentioned in three different channels this week. You can’t prompt your way to context you don’t have.Thor comes to you
We built Thor on a different premise: the AI should pay attention so you don’t have to. Instead of waiting in a chat window, Thor watches the actual flow of work — Slack threads, meetings, ticket activity, PRs, docs — and pings you when something matters. Not when anything happens. When something you would want to know about. That includes:- A meeting you were in just ended, and you’ve got three action items.
- A teammate said they’d follow up on something yesterday and haven’t.
- A product change you’re waiting just got blocked.
- A decision you cared about was made in a thread you missed.
- It’s Friday at 4pm and you struggle to remember what the team actually got done this week.
A chief of staff for your work
The mental model that fits best: a really good chief of staff. They sit in on every meeting. They keep tabs on every channel. They read every PR description. They notice when something stalls. They notice when a question went unanswered. They notice when a decision got made but nobody wrote it down. And — most importantly — they don’t wait to be asked. They DM you and say “hey, I think you wanted to know about this.” That’s Thor.A note on expectations
Thor’s good at this, but it’s not perfect. Sometimes Thor pings you about something you don’t care about. Sometimes Thor misses something you wish it had caught. When that happens, you tell Thor — in plain English — to back off on a topic, or to pay closer attention to one. The more you use Thor, the better it gets at picking the right moments. This is still early days. Some rough edges are expected.Reactive vs. proactive, in practice
Same task, two ways: Reactive AI:You: “What did we decide about pricing in last week’s leadership meeting?” → AI: searches transcripts, returns answer.
Thor, the morning after the meeting: “Heads up — three pricing decisions were made in yesterday’s leadership sync. Two need follow-up from your team. Want me to draft action items?”
Things Thor does without being asked
Out of the box, Thor will:- DM each meeting participant a recap of action items and decisions a few minutes after a meeting ends. See Meetings.
- Send you a daily standup with what you worked on yesterday. See Stand-ups.
- Send a Friday recap and a Monday kickoff. See Weekly Messages.
- Tell you when a task you care about gets blocked, completed, or reassigned. See Task Subscriptions.
- Follow up the next day if you haven’t closed out an action item from a meeting.
Hey
@Thor, let me know if anyone mentions the auth migration this week.Hey
@Thor, every Friday at 4pm, send me a list of customer-facing work I shipped this week — written for a non-technical audience.