Platform02

platform/autonomous-engine

Automation is easy. Automation you can undo is the product.

Anything can write to a marketplace API. The engineering is in what happens when the write half-succeeds, the channel rate-limits you at 2am, a supplier revises the document underneath a published claim, or somebody needs yesterday’s catalog back.

A unit of work, end to end
Admittedidempotency keyExecutedcheckpointedGatedcompliance + schemaEmitteddelta onlyDuplicate suppressedsame key, one writeResume from checkpointnot a restartHeldunsubstantiated claimBackoff, re-queued429 from channelrevert to any prior versionAPPEND-ONLY

Every transition is recorded · every write carries the version of the input, the agent and the rule set that produced it

0
Destructive writes

History is append-only. Reverting is a write, not a restore.

0%
Writes versioned

Input, agent and rule-set version recorded per write.

0.0%
Records re-sent daily

Deltas only. Unchanged records never touch a channel API.

0.00%
Runs settle

Acknowledged and reconciled against live channel state.

01Guarantees

Six properties, and what each one costs.

Every guarantee here is a trade. We would rather name the trade than describe the runtime as simply reliable.

01

Idempotent by key

Every unit of work carries a caller-supplied key. Replaying a run cannot produce a second write, so a retry after a network failure is always safe.

02

At-least-once, deduplicated

Delivery is at-least-once because exactly-once across a third-party API is a fiction. Duplicates are absorbed at the write boundary rather than wished away.

03

Versioned writes

Every write records the input version, the agent version and the rule-set version that produced it. A result is reproducible or it is not trusted.

04

Backpressure, not failure

When a channel rate-limits, work queues rather than erroring. Priority classes let a price correction overtake a bulk backfill.

05

Bounded blast radius

A run is scoped to a SKU set and a channel set. There is no operation whose failure mode is the whole catalog.

06

Reversible

Because history is append-only, any publish can be reverted to the exact prior state, including on channels that have no undo of their own.

The one we get asked about

At-least-once rather than exactly-once. Exactly-once delivery across a third-party API you do not control is not achievable — a channel can accept a write and fail to acknowledge it. What is achievable is making a duplicate harmless, which is what idempotency keys are for. Any vendor claiming exactly-once against a marketplace API is describing a wish.

02Lifecycle

Five states. One of them is a success that looks like a failure.

In this state now

0

units of work

Compliance and schema validation. Nothing proceeds that cannot be substantiated.

Work sitting at the gate is not backlog. It is exposure that was caught before it reached a customer, a regulator or a marketplace. A runtime with an empty gate queue is not performing well — it is not checking.

Priority classes under a rate limit
Price / stockclass P0Compliance fixclass P1Bulk backfillclass P3Rate budgetper channelChannelacceptedP3 YIELDS WHEN P0 IS WAITING

A price correction is not queued behind a bulk backfill · budget is allocated per class, not first-in-first-out

03Failure

What happens at 2am, specifically.

Uptime figures describe the good case. These are the six bad cases and the exact behaviour of each.

Channel returns 429

Work re-queues with backoff inside its priority class. Nothing is dropped, nothing is retried into a wall.

delayed

Channel returns 500

Retried under the same idempotency key, so a write that actually succeeded before the error does not duplicate.

retried

Agent times out

The run resumes from its last checkpoint rather than restarting the SKU from the beginning.

resumed

Compliance gate blocks

The run halts at the gate. This is a successful outcome, not an error — the exposure was caught before publication.

held

Source document revised

Affected attributes are invalidated and re-derived. Downstream emissions are re-queued automatically.

re-run

Operator reverts

The prior version is re-emitted to every affected channel from append-only history.

rolled back

Note that none of the six outcomes is “error, investigate tomorrow.” Every failure mode a marketplace integration actually encounters has a defined behaviour, because the ones without a defined behaviour are the ones that become a Monday morning.

Next

Run it against one channel, in dry-run.

Connect a single channel in read-and-simulate mode. You get the full diff of what would be written, the rate-limit projection, and the rollback plan — with nothing actually sent until you say so.

One channel · Nothing written · Full diff and rollback plan