Skip to main content

Workspace

Agent-native project management

Brief, run, review, and accept work in one place.

Keep the goal, task, conversation, artifact, worker, and decision on one shared board.

The work moves. The context falls behind.

The brief sits in one tool, the task in another, the run in a third, and the decision in a thread. Reviewers spend time finding the story before they can judge the result. MIT linked failed AI pilots to weak workflow integration and weak learning from prior work (MIT NANDA, 2025).

Example — fictional team and data. Not a live run or customer result.

Launch Operations · AES / Board

10:11

Ready0
Nothing queued
In progress0
No active runs
Approval1
AES-179
PR

Transfer customer settlement funds

Held by policy · P-12 · $80,000

Waiting on Priya Raman · 10:03

Complete1
AES-184
MC

Ship the governed customer import

Accepted

customer-import-v1.0.0 · 10:11

Meridian Freight's board at rest: AES-184 accepted, AES-179 held for a person.

Scattered work has no clear finish line.

Before

Which version is accepted? Who reviewed it? Is approval complete?

  • Salesforce: requirement
  • Notion: brief v3?
  • Slack: decision, in a thread
  • GitHub: artifact
  • Manual follow-ups, duplicated review, status reconciliation

After

One acceptance record.

  • Requirement, brief, decision, and artifact on one task
  • Reviewed ✓ Approved ✓ Accepted ✓
  • Completed work, durable record

One card carries the work.

  • Assign a real owner

    Give a person, agent, tool, or workflow a named step and keep its progress on the task.

  • Carry the brief forward

    The goal, constraints, comments, dependencies, and artifacts travel with the assignment.

  • Keep decisions with the task

    Handle questions, handoffs, exceptions, and approvals beside the work they affect.

Example — fictional team and data. Not a live run or customer result.

Plan

Live
Launch Operations · Goal tree09:24
  1. GoalLaunch the governed customer importOn trackMCMaya Chen · 09:00
    33%
  2. ProjectCustomer data launch1 taskActive
  3. AES-184Ship the governed customer importIn progressClaude Code · implementing PR #184
ThreadPinned
AES-184 · Ship the governed customer import
MC
Maya Chen09:31

Can we ship without copying customer data into the staging report?

Claude Code09:33

Yes. The staging report now carries aggregate counts only. The raw export stays behind P-12 sign-off.

Decision recorded09:34

Use aggregate counts in staging.

AES-184 · P-12 · Claude Code

Review the result with its history attached.

  • One home for each artifact

    Attach the spec, draft, diff, document, or output to the task and retain its source and revisions.

  • Acceptance stays separate from delivery

    A result can arrive without being accepted. Reviewers see the requirement, history, checks, and artifact before they decide.

Example — fictional team and data. Not a live run or customer result.

Review

Round 1
customer-data.ts+52rev 7
82 const report = createReport();
report.rows = rows;
report.export = exportRaw(rows);
83+report.count = rows.length;
84+report.byRegion = countBy(rows);
85+// raw export needs P-12 sign-off
86+report.export = null;
87+report.policy = "P-12";
88 return report;
  • Requirement · Staging reports carry aggregate counts, never raw customer rows.Satisfied
  • Tests passed · Aestus Review · 09:41

Historyspec rev 4diff rev 7review round 1

Follow one card from goal to acceptance.

Every step adds to the same work record.

Example — fictional team and data. Not a live run or customer result.

One card, six steps

AES-184
  1. 09:00

    Goal set

    Brief attached

    MCMaya Chen

    PM connects the goal, brief, repo, and owner.

    Customer import goal briefReady
  2. 09:08

    Spec drafted

    Planning agent drafting spec

    Aestus Planning

    Planning agent drafts the spec; PM redirects scope.

    Governed customer import specIn progress
  3. 09:24

    Code written

    Coding agent implementing

    Claude Code

    Coding agent starts with the full brief; nothing is re-explained.

    PR #184 · customer importIn progress
  4. 09:41

    Checks run

    Tests passed

    Aestus Review

    Tests run; the policy in force is recorded on the run.

    Import validation reportIn progress
  5. 10:03

    Reviewed & approved

    Held for approval

    PRPriya Raman

    Human gate · P-12

    Reviewer gets the diff, requirement, checks, and history.

    P-12 approval recordApproval
  6. 10:11

    Shipped & learned

    Accepted

    MCMaya Chen

    Outcome, cost, and acceptance are recorded for later comparison.

    customer-import-v1.0.0AcceptedComplete
09:0010:11Goal to acceptance in 71 minutes, on one record.

Three hidden costs of AI work end here.

  • Rebuilt context

    The objective and decisions stay attached through each handoff.

  • Review pileups

    Reviewers spend their time on the result, not on finding its history.

  • Coordination drag

    One task shows the current owner, artifact, review, and next decision.

The numbers this page owns.

First-pass acceptance rate
The share of results accepted without another production pass.
Request-to-completion cycle time
Elapsed time from the request to a recorded acceptance decision.

We already have a tracker with AI features.

Keep your tracker. Aestus adds the run, artifact, policy, review, cost, and acceptance record needed when people and agents share the work.

Watch a workflow go from goal to accepted completion.

Talk to us