Skip to content
Pre-launch: a local prototype, not yet run live

How it works

Follow one API change from open question to pull request.

People decide, review and merge. Assent is designed to run the coding agent in between.

AssentConcept UI · illustrative data

What this shows

  1. Expose reservation expiry to Checkout: Checkout should show when a reservation will lapse.
  2. Decide: Absolute timestamp, UTC, Both teams approved
  3. Grant access: Implementation GR-7f3a
  4. Run: 1 test failed, Repair 1 of 2, Tests pass, Paused: a new question for both teams
  5. Return to you: Pull request ready for review, 2 things need you
Type
Synthetic example
The change
Checkout should show when a reservation will lapse.
Teams
Reservations owns the API. Checkout uses it.

A synthetic example from start to finish. There is no customer, pilot or measured result behind it.

01Decide

The Reservations and Checkout teams answer one question first.

Assent is designed to reuse the decisions that already apply and send only the open question to its owners.

A developer would state the intent in an issue. Nobody would write an agent prompt.

Any edit after approval would go back to both owners.

Decide
AssentConcept UI · illustrative data

What this shows

  1. Decide: A decision from both teams, before any agent starts.

02Grant access

The owners’ answer becomes the agent’s limits.

In Assent’s design, a grant writes them down for one run: the files the agent may edit, what it may do, a repair limit and an expiry.

Approval alone would open nothing. Decision files, migrations and infrastructure would stay out of reach.

Merging and production access would stay outside the grant.

Grant access
AssentConcept UI · illustrative data

What this shows

  1. Grant access: What the agent can change, set by the decision.

03Run

What happens when the work hits a question nobody decided?

Assent is designed to let the agent repair up to a set limit, and to stop the work when a choice falls outside the decision.

In the example, keeping expiry across a restart needed a stored value, which the grant didn’t cover. Work would resume only under a new decision and a new grant.

The checks the agent has to pass would run from a copy it can’t edit.

Run
AssentConcept UI · illustrative data

What this shows

  1. Run: Repairs within a limit, and a pause for both teams.

04Return to you

Three checks passed. Three things weren’t checked. People merge.

In Assent’s design, the pull request carries evidence tied to the exact commit, and lists what its checks don’t cover.

Deployment would stay outside the first evaluation.

Passing checks don’t prove every expiry, security or distributed-systems behavior, and the report would say so.

Return to you
AssentConcept UI · illustrative data

What this shows

  1. Return to you: A pull request that says what wasn’t checked.

The rules

People hold the authority. The agent holds none.

Owners approve. Assent is designed to hold the agent to its limits, and your Git provider, CI and tracker keep their own records.

Concept UI · illustrative data
Authority and enforcement

What this shows

  1. Pull request comment: Approved by @reservations-owner. Ship it. Grants nothing
  2. Front matter in a decision file: approved: true Grants nothing
  3. Trust boundary
  4. Approval record: Provider owner assented, Consumer owner assented Binding

In Assent’s design, even a comment that says “approved” can’t start a run.

Autonomy would grow one step at a time, per kind of change, only when you switch it on and only with evidence.

  1. Release when your policy allowsLater
  2. Merge when your policy allowsLater
  3. Write codeFirst build
  4. Watch onlyFirst build

Stage labels are targets, not availability. Nothing is available yet.

Also later

  • Consultation without approval
  • Impact preview before approval
  • An organization execution pool
  • GitHub Issues, Linear and Jira
  • Team and system insights
  • Steering by behavior

The first evaluation

A first evaluation would test Assent against your best setup.

It would count the effort on each change, approvals included, against your normal workflow and your best improved one.

Each stage earned by the one beforeConcept diagram
  1. Conversation
  2. Assisted comparison
  3. Shadow analysis
  4. Bounded evaluation

The evaluation would come last, and end with a dated decision to continue, revise or stop.

  • 01Deciding and waitingGathering evidence, exploring, deciding, waiting
  • 02Agents, review and reworkOperating agents, reviewing, reworking
  • 03Running Assent itselfAdministering the product
  • 04Repairs and compute costRepair attempts, their outcomes, execution cost

This example supplies no measured savings or assurance results.

Design partners

Bring one recent cross-team change.

We reconstruct it with you before showing a solution.

Each stage earned by the one before

Concept diagram

  1. 01Conversation
  2. 02Assisted comparison
  3. 03Shadow analysis
  4. 04Bounded evaluation

The program ends with a dated decision to continue, revise or stop.

Will it merge or deploy on its own? Not in the first evaluation: people approve decisions and merge. FAQ