Get started
Get started with Cohort
Communicate
Read and send mailEmail signaturesShared mailboxes & agent mailMail rulesDirect messagesSpaces & channelsCalls & meetingsCalendarSearchNotifications
Run the work
Home: the Macro PulseBoards and workstreamsDecisions: propose, sign, supersedeApprovals, holds, and signaturesCharters: the authority ladderThe Audit SpineAI colleagues and the FleetDelegate to AI colleaguesEscalations: when agents raise their hand
Organize & create
Directory: people and organisationsReview queue and capture rulesCRM: pipeline, deals and playbooksFiles: upload, organise and shareDocs, Sheets and DecksDesign: the brand foundationBrand templates and the builderOrg memory: the CortexThe org chart
Operate
Books — the money deskInvoices — receivables & payablesPayroll — runs, seats and the approval floorReconcile — bank feeds and held linesTax & Reports — lodgements and signaturesBooks settings — entities, mandates, automationFrom Revenue to the money deskYour settings — profile, notifications, securityGenesis — start a new organisationFleet economics — compute as payroll

Decisions: propose, sign, supersede

Every consequential call — made by a human or an agent — is a row on the decision ledger at Org → Audit Spine. The ledger is append-only: decisions are never edited in place, they are signed, superseded, or reversed, and the full "why" travels with each one. Humans and agents write to the same ledger, so there is exactly one record of what the company decided.

The Audit Spine page: the decisions ledger as a day-grouped timeline with facet filters and a New decision button
The decision ledger — newest first, grouped by day

Read a decision row

Collapsed, each row is a git-log line: a presence dot (olive for an agent, ink for a human), the title with a mono timestamp, the proposer's avatar, an act-type gate chip (Auto · Auto-notify · Approval-gated · Propose → approve · Human-gated), the Cortex version it relied on, and a reversibility flag.

Click a row to expand the full record: reasoning, the options considered, evidence, what would change the call, the derived-from clause (linked to its source authority in Cortex), the signature list, and the discussion thread.

Status maps to an outcome chip:

| Status | Chip | Meaning | | --- | --- | --- | | proposed | Escalated | Raised, awaiting endorsement | | adopted | Mandate | Within charter, enacted | | superseded | Overridden | Replaced by a successor | | reversed | Overridden | Undone within its window |

Propose a decision

  1. Open the form

    Click New decision above the timeline. Proposing requires EDITOR access — a proposal enters the org's permanent record.

  2. State the call

    Title ("What was decided?") and rationale ("Why this call, why now?") lead. Add a tag (launch · cost · policy · safety) for the facet filters.

  3. Declare reversibility

    Pick the window: Reversible, Within 7 days, Within 30 days, Within Q, or Standing. Decisions are reversible by default; Standing means there is no undo.

  4. Attach the thinking

    Options considered and evidence go in one per line. "What would change this" records the trigger that should reopen the call. The derived-from clause cites the charter or policy line the decision rests on.

Proposals scoped to a Space also post a system message into that Space's stream, so the room sees the call where the conversation happened.

Duplicate proposals are refused

Re-submitting a proposal with the same title while your previous one is still open is treated as a retry, not a new decision — sign, supersede, or adjust the existing row instead.

An expanded decision row showing rationale, options, evidence, derived-from clause, signatures, and the discussion thread
The expanded record — the full why, on the record

Sign to endorse

Signing appends your attestation — your member key and a server-stamped time — to the decision's signature list. Signatures are append-only and idempotent: you can't sign twice, and nobody can sign as you (the signer is always resolved server-side from the acting session). A superseded decision can no longer be signed.

You can sign from the expanded row here, or straight from the decision's card in the Home needs-you queue — both paths reach the same audited write, and the same is true when an AI colleague endorses a decision.

Supersede, adjust, or reverse

  • Supersede — replace a decision with a successor. The form pre-fills from the predecessor; on submit the new row is created and the old one flips to superseded atomically, with the successor pinned to it. The room is told what replaced what. Superseded rows become read-only.
  • Ask to adjust — request changes without killing the proposal. Your feedback lands in the decision's discussion thread flagged as an adjustment request, and the decision stays open while the proposer revises. You can @-mention colleagues in the ask.
  • Reverse — undo a reversible decision within its declared window. Only proposed or adopted decisions can be reversed; the row flips to reversed on the record rather than disappearing.

Adoption — the proposed → adopted transition that turns a proposal into a mandate — is exercised through the agent plane's governance-gated decision.adopt method, and every transition lands on the spine either way.

Discuss without mutating

The discussion thread under each decision persists comments alongside the row — the decision itself is never edited. Every comment also chains a spine event, so the rationale trail stays complete. Deep links work across the app: a decision row anywhere (Home, a Space's Decisions tab, a mention) lands here with the row expanded and scrolled into view.

The ledger is live: an agent's decision.proposed or decision.signed appears without a reload.