An outage shouldn’t stopyour whole team.

Your engineers should be building great things, not fixing outages or running cloud ops.Ace is a team of AI agents that does that work for you.

In production with

AcmeInvestigations / ACE-842

Investigatingorders-platform

Orders stalled after a schema change

orders-api / request trace8.4 s

02 s4 s6 s8 s

HTTP request

8.4 s

Authentication

38 ms

Acquire connection

42 ms

SELECT orders

8.1 s

Publish outbox

waiting

Lock chain Downstream impact

Amazon RDSpg_locks14:04:12E1

A migration waits on a lock held by reporting

Infrastructure agent

Datadogrequest trace14:04:19E2

Order delivery waits behind the same session

Telemetry agent

GitHubdeployment 84214:04:31E3

The migration runner retries without a timeout

Code agent

db/migrations/runner.tsPR #843

24await db.transaction(async (tx) => {

25+ await tx.sql`SET LOCAL lock_timeout = '2s'`;

26 await applyMigration(tx, migration);

28− retry: untilSuccess()

28+ retry: boundedBackoff({ attempts: 3 })

3 checks passedApproval required

Orders API · p99184 ms

14:03Recovery applied14:12

Lock queue clearedOutbox caught up

Saved to system memory

Agents investigating

A whole team. On the same case.

Infrastructure agent

Checks the resources behind the failure and the services that depend on them.

Telemetry agent

Follows the failing request and shares the trace that sends the others to the right service.

Code agent

Tests recent changes against the findings, then prepares the patch and a regression test.

Recovery agent

Rechecks the original failure and downstream impact, and reopens the work if recovery fails.

Know what depends on
the failing service.

Ace maps the resources and dependencies in your infrastructure. Its agents use that graph to follow a failure across services, even when the alert only names one.

System map

Opening system map

Review the fix
where you ship code.

Ace opens a pull request with the patch, the evidence behind it and a test for the failure. Your engineers review and merge in GitHub.

acme / orders-platform

Bound migration locks and retries #843

Openace wants to merge 2 commits into main

db/migrations/runner.ts+8 −2

24await db.transaction(async (tx) => {

25+ await tx.sql`SET LOCAL lock_timeout = '2s'`;

26 await applyMigration(tx, migration);

27});

28− retry: untilSuccess()

28+ retry: boundedBackoff({ attempts: 3 })

3 checks passed

Fix pull request
A minimal code or infrastructure-as-code repair, tested in an isolated sandbox and opened for review.
Revert pull request
An exact reversal of the change behind the incident, ready to merge.
Rollback
Redeploys a known-healthy revision through a workflow your administrator has confirmed.
Kubernetes changes
Restart, scale or patch workloads in the namespaces you grant, under the same approvals.

Memory that stays
with your system.

Ace keeps the failure patterns it has seen, the hypotheses it ruled out and the fixes that held, along with your engineers’ feedback. The next investigation starts from all of it.

Patterns
A schema migration waits behind an open transaction
Recognized the next time the same shape appears in your system.
Ruled out
Database CPU was 19%. Capacity was not the cause.
Dead ends and false alarms stay closed, with the evidence that closed them.
Fixes that held
Lock timeout and bounded retries, recovery verified
Reused for similar changes. Fixes that failed are remembered too.

Delegate more
as Ace earns it.

Start with investigations. Approve each change while you review the results. Turn on automatic actions only where Ace has already proven itself, within limits your team sets.

How earned autonomy works

Autonomy policyorders-platform