Never forget a commitment your team made

People make promises in Slack, email, and meetings that vanish into the scroll. Projects get discussed for weeks without progress. Attest Manager watches everything — Slack, email, meeting notes, GitHub, Jira, Google Docs — extracts commitments, detects stalls, and drafts supportive check-ins you review before sending. For any team: engineering, sales, product, ops, support, finance.

What it detects
Commitments, stalls, and wins — across every team
LLM-powered commitment extraction + rule-based anomaly detection. Every signal is transparent and shows its source.

Missed commitments

Someone promised a deliverable in Slack, email, or a meeting — and the deadline passed without delivery.

"Alice said she'd send the pricing deck by Friday in #sales. It's Monday and no doc activity."

Project stall

A feature or initiative keeps getting discussed in meetings and Slack but nothing tangible is shipping — no docs, PRs, or ticket movement.

"'Payments redesign' discussed in 4 consecutive standups but only 1 PR opened in 3 weeks."

Meeting loop

The same action item keeps appearing in meeting notes, week after week. Someone's stuck or it keeps falling off the priority list.

"'Update pricing page' has been assigned to Sarah in 3 consecutive meetings."

Blocker persistence

Someone said they're blocked days ago and nothing has changed. The blocker is still there.

"Bob mentioned being 'waiting on legal review' 4 days ago. No follow-up activity."

Communication-delivery gap

Lots of discussion about a topic across channels, but no corresponding output — no docs updated, no tickets moved, no artifacts produced.

"'Q3 roadmap' mentioned 18 times in Slack this week but the Google Doc hasn't been edited since March 12."

Delivery recognition

Someone shipped something notable: a big deliverable landed, a deal closed, a feature launched, a support backlog cleared. Time to say thanks.

"Grace closed 8 support tickets and published the new FAQ page this week."
How it works
From signal to check-in, with you in control
1

Connect your data sources

Slack, Gmail, Google Docs, GitHub/GitLab, Jira/Linear, Confluence, Notion, meeting notes. 30 connectors, all read-only, periodic polling. Attest ingests everything as provenanced claims.

2

Commitments are extracted automatically

LLM reads messages and meeting notes, extracts "I'll do X by Y" promises. Each becomes a tracked claim linked to the person, the deliverable, and the deadline. No manual entry.

3

Signals surface in your review queue

Missed deadlines, stalling projects, meeting loops, persistent blockers, and shipping wins all appear with full context: who said what, where, when, and what's happened since.

4

Draft messages you can edit or dismiss

The system drafts a supportive Slack DM. You edit it, approve it, or dismiss it. Nothing is ever sent without your explicit approval. The tone is always encouraging.

5

Knowledge compounds

Every signal, every check-in, every outcome becomes a claim in the knowledge graph. Over time, the manager learns which signals matter and which are noise.

Design principles
Built for trust, not control

Human-in-the-loop always

The daemon never sends a message. It generates drafts. You review every single one. Edit, approve, or dismiss — your judgment, not an algorithm's.

Encouraging, never accusatory

Draft tone is "checking in" not "I noticed you're not working." Configurable: supportive (default), direct, or casual. Built to strengthen relationships.

Celebrate wins, not just flag problems

The shipping streak detector recognizes when someone is on a roll. Great managers acknowledge great work. The system helps you do that consistently.

Transparent baselines

Every signal shows its math: "3 days vs normal 0.5 day gap." No black box ML. You can see exactly why a signal fired and decide if it's real.

Read-only connectors

Slack, GitHub, Jira connectors only read. The only write operation is sending an approved DM — and that requires explicit action from you.

Claims, not summaries

Activity isn't flattened into summaries. It's stored as provenance-tracked claims. You can drill into any signal and see the exact commits, messages, and tickets.

Who it's for
Any manager with a team and too many channels

Engineering

Track PRs, deploys, ticket velocity. Spot when a feature is stalling or a team member is blocked.

Sales

Follow up on proposals promised to prospects. Catch deals going quiet. Recognize when someone closes big.

Product

Detect spec reviews stuck in meeting loops. Track when roadmap items are discussed but not started.

Support

Spot escalation patterns. Recognize agents clearing backlogs. Track response commitments to customers.

Ops / Finance

Track deliverable deadlines across teams. Catch quarterly planning items that keep slipping.

Any team

If your team communicates via Slack, email, or meetings — commitments are being made and lost every day.

For your coding agent
12 MCP tools
Setup & control
team_setup(team_name, members_json)        # configure team members
team_add_member(team, email, slack_id)     # add a member
team_configure(team, slack_token, tone)     # connect Slack, set tone
team_monitor_enable(team)                   # start background daemon
team_monitor_disable()                      # stop daemon
team_monitor_status()                       # check cycle count, pending drafts
Intelligence & drafts
team_dashboard(team, period="7d")           # per-member activity summary
team_member_detail(email)                   # deep-dive on one person
team_check_now(team)                         # run detection immediately
team_review_queue()                          # list pending draft messages
team_edit_draft(draft_id, new_text)          # edit before sending
team_send_draft(draft_id)                    # approve and send via Slack
Part of the Attest platform

Attest Manager runs on the same claim-native database as Attest Brain. Same provenance, same confidence scoring, same knowledge graph — applied to team activity.

Attest Brain Quick Start GitHub