Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.janet.ai/llms.txt

Use this file to discover all available pages before exploring further.

Want a JanetClaw on your team? Email adhit@janet.ai. We ship the Mac Mini, set everything up, onboard JanetClaw onto your tools (Janet AI, Slack, Gmail, Calendar, Drive, GitHub, Claude Code), and run a short training so your team knows how to use it. Plug it in and it’s running.
JanetClaw is your 24/7 AI project manager. It runs on a Mac Mini we ship to your office, listens to every channel of communication your team uses, and handles the project management work nobody wants to do — capturing tasks as tickets, updating tickets when more context comes in, scheduling meetings, sending reminders, even drafting PRs to resolve tickets — all fully in the background, without anyone having to ask. Think of it like OpenClaw, but tuned exclusively for Janet AI. It’s a teammate, not a tool.

What It Does

Out of the box, every JanetClaw does the following 24/7:
ToolWhat it does
janet.ai workspaceAlready wired to manage your Janet projects and tickets perfectly. Every ticket has all relevant context from your team and customer communications auto-attached, and gets updated as you communicate progress. You can ask JanetClaw to take any action across your whole workspace.
SlackAlways listening to your conversations and turning them into ticket creations and updates, with the full thread attached as context. Also your main channel to talk to your JanetClaw - ask it anything and it’ll look across every information source it has to synthesize whatever project info you need.
GmailAlways listening to every inbox connected. Any email — customer support, meeting notes, anything actionable — becomes a new ticket or an update to an existing one with the full thread attached as context. Scheduling emails can be handled end-to-end too.
Google CalendarAutomatically finds available slots for scheduling your calls and sends you meeting prep every morning, pulling from your past communications with each attendee and the web.
Google DriveSearch, share, move, rename, manage permissions upon request.
Product DocsReads your product documentation and answers external sender questions in Slack and Gmail threads and replies whenever the docs already cover the answer.
Coding TasksKicks off draft PRs on tickets using Claude Code via the Janet MCP when asked to implement a task — clone, code, push, link.
RemindersTell JanetClaw to remind you about any “check back later” moments — follow-ups, nudges, deadlines.
Activity FeedEvery action JanetClaw takes in the background posts to your JanetClaw page in your Janet AI dashboard for your team’s observability at all times.
Today
Created ticket from Slack11:42 AM
Created ENG-218 “Customer hitting API rate limits when bulk-importing” from a #support thread. Tagged support, assigned to @priya.
Coding task started11:08 AM
Kicked off a draft PR on ENG-215 “Investigate intermittent webhook delivery timeouts” via the Janet MCP.
Coding task started10:31 AM
Kicked off a draft PR on ENG-187 “Add audit log export to org settings” via the Janet MCP.
Updated ticket from Slack10:14 AM
Added context to ENG-212 “SAML SSO for enterprise” — attached a #sales thread where two more enterprise prospects asked about it.
Scheduled meeting9:46 AM
Booked “Q3 roadmap sync” with founder@partner.io for Thu 2:00 PM PT, replied to the email thread with the invite, and DM’d @adhit a Slack prep brief with the latest status on every ticket tied to past conversations with them.
Yesterday
Created ticket from email4:22 PM
Created ENG-212 “SAML SSO for enterprise” from an email by ops@acme.com and added it to the Enterprise Q3 objective.
Reminder fired3:00 PM
Nudged @sarah: ENG-204 “Add audit log retention policy” has been in code review for 3 days.
Added context to ticket2:38 PM
Attached a #support thread with a customer repro to ENG-201 “Webhook signing inconsistency between v1 and v2”.
Reminder created1:54 PM
Will follow up in #design-review if Adhit’s RFC thread sits unanswered for 4 hours.

A Teammate That Learns

JanetClaw learns and remembers. Tell it how you want something done — what counts as a bug vs. a feature request, which channel maps to which project, who owns what — and it does it that way from then on. Give it feedback in a Slack DM (“assign all frontend issues to me”, “don’t ping me about these kinds of issues”, “make tickets for every customer email in this project”) and it updates its own playbook with the new flows. The more you use it, the more it behaves like a member of your team.

Frequently Added Flows

A few of the most common things teams teach their JanetClaw. Each one started as a single Slack DM to JanetClaw — JanetClaw turned it into a permanent workflow in its memory on its own.
“When you see an email with meeting notes from notetaker@yourcompany.com, parse every action item in the summary. For EACH action item, FIRST semantic-search our existing tickets — if a matching ticket already exists, attach the meeting-note email to it as context and fold any new details into the description. ONLY create a new ticket if the search returns nothing relevant. Never duplicate. Every time.”
For every action item, JanetClaw runs search_tickets_semantic against your workspace before doing anything else. If a matching ticket exists, it attaches the meeting-note email as context and updates the description with any new details — no duplicate ticket gets created. Only when the search comes back empty does it open a new ticket, with the email already attached.
“Every morning at 9am, send me a briefing on all my meetings for the day. For external calls, look up email threads and Slack threads with the attendees, search our workspace for any tickets related to past conversations with them, and web-search them. Three lines per meeting: intent, what they’ve said before, and where every related ticket stands today — current status, latest update, expected timeline.”
A 9am heartbeat reads your calendar, pulls Gmail history and any Slack threads with each attendee, semantic-searches your janet.ai workspace for tickets tied to past conversations with the customer, web-searches external attendees and their company, and DMs you a per-meeting brief — including a status line per related ticket (e.g. “ENG-212 SAML SSO — In Progress, PR #418 in review, ETA next sprint”) so you walk in knowing exactly what to say.
“Anytime an external sender posts in any Slack channel, remind us to follow up by pinging again in that channel if no internal teammate has replied within 24 hours.”
Every external Slack message in a watched channel queues a 24h reminder. If a teammate replies, the reminder clears itself. If not, JanetClaw posts a nudge in the thread so the question doesn’t get lost.
“Whenever a scheduling email comes in — a new request, a reschedule, or a cancellation — check my calendar, propose the closest valid slot (or confirm/cancel), send or update the Google Calendar invite, and reply to the thread confirming the change. If anything is ambiguous, DM me on Slack with the proposed action before sending.”
JanetClaw classifies each scheduling email as new / reschedule / cancel, finds a slot that respects work hours and avoids conflicts on the inbox owner’s calendar, creates or updates the Google Calendar invite (with a Meet link), and replies on the thread with the confirmed change. Cancellations clear the invite and acknowledge the thread. Anything genuinely ambiguous gets staged as a Slack DM for one-click approval before any email goes out.
“Whenever a customer emails about a bug, broken behavior, or a product question, semantic-search our resolved tickets in janet.ai for similar issues we’ve already shipped fixes for — their descriptions and comments have the implementation details — and also check any in-progress tickets covering the same area. Combine that with our past replies on similar threads, draft a response in my voice, attach the email to the right ticket, and DM me on Slack with the draft so I can review and send.”
JanetClaw runs a semantic search across your resolved tickets (where the description, comments, and linked PRs have the actual implementation context) and any in-progress tickets touching the same area, blends in your prior outbound replies on the same topic, and drafts a response in your voice. It attaches the email to the matching ticket and DMs you on Slack with the draft + a one-click “send as-is” option — including a short “based on” line listing the resolved/in-progress tickets it pulled from so you can sanity-check the answer.
“If you’re not confident in the draft reply for a customer email, open the on-call sheet in Google Drive (the one named ‘On-Call Rotation’ in our Engineering folder), figure out who’s on call right now, and Slack DM them a one-paragraph thread summary plus the single specific question you need answered. When they DM back a one-liner, turn it into a polished customer-facing email in my voice — no internal ticket links — and DM me the draft on Slack alongside any related janet.ai tickets so I can sanity-check before sending.”
JanetClaw opens the on-call rotation sheet in your Google Drive, parses the schedule to find whoever’s covering the current shift, and Slack DMs them with a tight thread summary + one focused question. Their one-line Slack reply gets rewritten into a professional, customer-safe email in your voice — no internal ticket links in the body — and lands in your Slack DMs as a ready-to-send draft, with the related janet.ai tickets surfaced in the Slack message so you can sanity-check before sending. Nothing goes out to the customer until you approve it.

Get a JanetClaw

Email adhit@janet.ai — we’ll get you set up.