AN AI PROGRAM MANAGER IN SLACK

Give your team its focus back.

Coordinated is a shared record of your decisions, guardrails and plan, so people and agents work from what has been agreed, without you having to repeat every decision or chase every update.

Your people read it on Monday. We are building the read contract your agents will use before they act.

Every message says why it was sent and what it cost you.

CoordinatedAPP

Staging may slip: Thu Oct 1 → Sat Oct 10

Maya's forecast, not yet a change. Nadia decides.

Your announcement (Thu Oct 15) needs ten calendar days after staging; Sat Oct 10 leaves five.

Your campaign spend locks Tue Oct 6, 5:00 PM.

You
asked
Marco
informed · his Acme condition still holds
Nadia
decides

Why you got this: the launch plan links your announcement to staging. This card updates in place; when Nadia decides, it reads “Decided Tue Oct 6, 9:05 AM”.

Reads only what you hand it, what is posted as artifacts, or threads where it is tagged

One card per commitment per person, updated in place

Built for 50 to 300-person software companies

Memory tells you what was said. The record tells you what was agreed.

01 / How it works

One change propagates through one record.

01 / 03

Hand it what you already wrote.

Coordinated proposes the initiative, people, commitments and dependencies, each with the sentence it came from. The operator reviews one draft.

02 / 03

Owners accept, condition, or decline.

Nothing is a commitment until a person answers. Conditions and proposed dates stay attached to the terms they qualify.

03 / 03

The review opens settled.

When a date moves, the people it reaches are named before it moves. Cards update in place, answers return, and the terms move to v2.

01 / 03
Illustration: Converging galaxy tails show a change reaching the people who depend on it.
Atlas launchv1
Staging may slip

Sarah · asked

Marco · informed · condition holds

Source sentences remain attached to every proposed term.

02 / Sample record

In the pilot today

A sample record. Six moments when it matters.

See all six product cards

Atlas, Acme and the people below are fictional examples, separate from the Lantern simulator fixture used in the comparison. In the first pilots our team delivers the cards and briefs by hand inside your Slack, under the same rules.

01 · A date slips

Fictional example · Atlas

“Staging slipped to Saturday. Who needs to know before Sarah's spend locks Tuesday at 5?”

Sarah is asked: her announcement needs ten calendar days after staging and Sat Oct 10 leaves five. Marco is informed: his Acme condition still holds by one business day. Nadia decides; the forecast is on record either way.

  • Launch plan ¶7 · ten days after staging
  • Delivery notes, line 11 · Acme rollout Thu Oct 15
  • Spend lock · Tue Oct 6, 5:00 PM

Decided Tue Oct 6, 9:05 AM · 1 asked, answered · 1 informed · 0 pending · terms v2

02 · A scope changes

Fictional example · Atlas

“Nadia removed self-serve import on Wed Oct 7. Which promises now need an answer?”

Sarah is asked because the launch announcement names the import. Marco is asked because the Acme rollout includes it. Dana is informed because the support guide no longer needs the import path.

  • Launch plan ¶4 · self-serve import in scope
  • Acme notes, line 8 · import workflow Thu Oct 15
  • Support brief ¶3 · guide due Mon Oct 12

Decided Thu Oct 8, 11:20 AM · 2 asked, answered · 1 informed · 0 pending · scope v2

03 · An owner leaves

Fictional example · Atlas

“Maya is out through Fri Oct 16. Who owns the staging decision due Thu Oct 8?”

Nadia is asked to name temporary authority. Josh is informed because the launch cannot move without that decision. No replacement is inferred; ownership stays unseen until a person accepts it.

  • Review notes ¶6 · Maya owns staging readiness
  • Atlas terms v2 · decision due Thu Oct 8, 2:00 PM
  • Team update · Maya unavailable through Fri Oct 16

Recorded Wed Oct 7, 4:40 PM · Josh accepted interim authority · 0 pending · ownership v2

04 · A promise is made to a customer

Fictional example · Atlas

“Marco promised Acme a Thu Oct 15 rollout. Does the current plan support it?”

Marco is asked to confirm the condition. Nadia decides whether it becomes part of the Atlas terms. Sarah is informed because her announcement uses the same rollout date.

  • Acme thread · rollout Thu Oct 15
  • Launch plan ¶7 · staging Sat Oct 10
  • Announcement brief · publication Thu Oct 15, 8:00 AM

Accepted Mon Oct 5, 2:10 PM · customer term recorded · Marco owns the update · terms v2

05 · An assumption comes due

Fictional example · Atlas

“The rollout assumes legal approval by Mon Oct 12. What happens if it is not in?”

Nadia is asked to decide before the approval window closes. Sarah and Marco are informed because their Thu Oct 15 work depends on the answer. The record labels legal approval unseen until someone posts it.

  • Atlas assumptions ¶2 · legal approval Mon Oct 12
  • Campaign brief · spend starts Tue Oct 13, 9:00 AM
  • Acme notes · rollout Thu Oct 15

Answered Fri Oct 9, 3:30 PM · condition added · review Mon Oct 12, 9:00 AM · 0 pending

06 · A review starts

Fictional example · Atlas

“What is already settled before the Tue Oct 13 review?”

The staging forecast is decided, the import is out of scope, Marco's customer condition is recorded, and every affected owner has answered. The brief opens with the remaining decision, not a reconstruction.

  • Atlas terms v2 · accepted Thu Oct 8, 11:20 AM
  • Owner responses · through Mon Oct 12, 4:15 PM
  • Review brief · generated Tue Oct 13, 8:00 AM

Review opened Tue Oct 13, 9:00 AM · 3 changes settled · 0 decisions reconstructed

03 / Attention

Measured on the simulator fixture

What coordination costs a team, with and without us

One launch: nine people, seven weeks, two material changes, one review. The cards, notes and posts in Coordinated's column are counted from our simulator's event log. The minutes, the meetings and everything in the "today" column are assumptions you can change, including the possibility that Coordinated adds work. Savings can be negative, and the table says when.

Team hours on coordination over the launch
In meetingsMessages and threadsCoordinated's own work

"Meetings smaller and shorter" keeps the first two status meetings, then a 15-minute sync with four people each week, a 30-minute review with four people, and a 15-minute discussion with three people per change. Nothing here is a customer result.

Change the assumptions
The launch
Today's workflow
Coordinated's own work
Rates
Person by person: what Coordinated sent in seven weeks

Every card is logged against the person who received it, with the reason it was sent. Counts are measured on the fixture (33 cards, 9 notes, seven weeks, nine people); reading minutes are estimates at the reading speed set above. The pilot readout uses this ledger format; its figures come from that pilot.

Thirty-three cards needed an answer, nine settlement notes needed reading, and every person read twelve channel posts, seven of them Monday pictures. The sponsor and the people who own nothing on the launch answered nothing. The week a person moves their own date is the week they hear from us: Maya moved hers, and five of her ten cards came that week. The Thursday brief derived "No meeting needed: nothing is waiting on a person in the room."

The question we ask at formation

Which recurring meeting does this launch retire, and from when?

The hours above come from that answer, and from nothing else. If every meeting stays, Coordinated adds work. The sponsor answers it on day one, the default being that the Monday picture replaces the status meeting from week three once it has been right for two, and the readout says whether the meeting stayed retired.

04 / Pilot

In the pilot today

Six weeks. One launch. Two responsibilities off your plate.

We form the record by interview: one card per commitment per person, the Monday picture, the Thursday brief and settlement notes. In the first pilots our team runs the desk by hand inside your Slack, under the same rules. Every hour of ours is logged in its own column.

Discuss a six-week pilot
ed@coordinatedai.ai

Chasing answers

Agree who needs to answer each change. Coordinated keeps their cards current and brings their answers back to the record.

Preparing the review

Start from the accepted decisions, conditions and open questions instead of reconstructing the launch from messages.

The readout

Measure effort to keep one decision ready, decisions before their deadline, responsibilities stopped, and meetings retired.

Ask us for the number

See what the readout measures →

Once we agree the scope, send one document with a real date; the first card goes out from it.

Optional, after scope is agreed

Share the agreed launch document.

We read it and reply.

05 / Trust

The boundary is visible.

What it reads

Documents you hand it, artifacts posted in claimed channels, and threads where it is tagged.

What it never reads

Channels on its own, direct messages, or meetings. It labels what it cannot see.

What a channel sees

Counts and state, never restricted terms. Each person receives only the card that belongs to them.

What we are building

Three agents, one record.

The design gives the driver the work of preparing decisions, the verifier the work of checking claims, and the aligner the work of comparing accepted terms with the sponsor's intended outcome. These agents are planned architecture, not available pilot automation.

Accepted, supported, or unknown.

We are building a verifier that will mark every claim as one of three things: accepted by a named person, supported by current evidence with the period it covers, or unknown. Nothing on the page or in a message is presented as true because someone accepted it.

Guardrails as code.

We plan deterministic checks over each proposed message and change. The checks will return a refusal when a rule is not met, with a person resolving what the record cannot settle.

A read contract for agents.

We are building a read contract so agents will be able to ask who decides, what was agreed and what changed. Only people will be able to change the accepted terms.

The record holds pointers to your documents, with the version and the person who decides each, and keeps the evidence you hand it under your retention policy.

06 / Start

One document. One change. One review.

  1. 01

    Agree the scope of one six-week pilot, then send one document with a real date.

  2. 02

    Review one draft; send one card per person.

  3. 03

    Work through one change before the next review.

Discuss a six-week pilot

Your reading receipt

What this page put on record.

We plan to send each person a receipt like this every quarter: what it asked, why, and what it cost.

This receipt updates as sections enter view. It stays in this browser and is not sent anywhere.

CoordinatedReader receipt
  • Homepage opened