Production incident
Any severity
Raise through the console or the on-call number in your runbook. Severity 1 pages the duty engineer immediately and opens a bridge.
- Named incident commander
- Written post-incident review
- Cover agreed per engagement
Help & support
Support
Documentation and reference live in the developer portal. Everything else — incidents, architecture, escalation — comes through the channels below.
Getting help
Four ways in, depending on what has happened
Technical support
Business hours, per region
Configuration, integration behaviour, unexpected results and performance questions. Raised in the console with the trace id attached.
- In-console ticketing
- Trace-linked cases
- Regional coverage
Architecture and change
Your named architect
Anything about how the estate should be built or changed: capacity, routing policy, evaluation design, agent envelopes, new use cases.
- Named contact
- Quarterly estate review
- Change planning
Documentation and reference
Developer portal
API reference, SDKs, CLI, Terraform provider, migration guides and code samples. Access is included with any paid subscription.
- Subscription required
- Versioned per release
- Migration guides
Severity
What each level means and what we commit to
Severity is set by impact, not by how the ticket is worded. If you believe a case is mis-classified, say so on the ticket and it is re-triaged — we would rather over-classify than argue.
| Severity | Definition |
|---|---|
| Sev 1 | Production decision path unavailable, or producing incorrect outcomes |
| Sev 2 | Significant degradation with a workaround, or a single workload down |
| Sev 3 | Component impaired, no material business impact |
| Sev 4 | Question, guidance, or feature request |
On response times. This page previously published a table of contractual response targets across every tier. We are a small team and we are not going to publish a 24×7 commitment we cannot yet staff.
What we will do is agree response and escalation terms in writing as part of an engagement, sized to what the estate actually needs and to what we can actually honour. If round-the-clock cover is a hard requirement for you, raise it in the first conversation rather than the contract review — it is a real constraint on us today and you should know that before you invest time.
Every Sev 1 receives a written post-incident review including the trace evidence, the contributing causes and the actions taken. That one is not tier-dependent and is shared whether or not you ask.
Self-serve
Before you raise a ticket
Common questions
Answers we give often
Why do I need a subscription to read the documentation?
The developer portal is part of the product rather than marketing material. It carries the API reference, SDKs, the Terraform provider, migration guides and working samples against a live environment, and it is versioned against the platform release your estate runs. Access is included with every paid tier, including Starter once a workspace is created.
Something changed in the output and nothing changed in my code.
Check the route in the trace. A capability-class route can resolve to a different model when a provider deprecates one or when an evaluation gate rejects a version. Every routing decision is recorded with its reason — that is the first place to look, and it resolves most of these cases without a ticket.
How do I get a decision explained to an auditor or a customer?
Use the audit pack export from the trace store. Filter to the decisions in scope and export; the pack contains the trigger, the evidence with record-level lineage, the constraint checks, the policy gate, the model versions and the outcome. It is designed to be handed over as-is.
Our evaluation suite is blocking a promotion we need.
That is the suite working. The override path exists and runs through your own risk owner rather than through us — we can tell you what failed and why, and we will not promote a version over a failing hard gate on your behalf.
Can we get support in our own language and time zone?
Support is in English today. We are a small team in a small number of time zones, so cover outside them is agreed per engagement rather than offered as standard — ask before you assume it. Sovereign estates with cleared-personnel requirements are supported by the on-site team for that programme.
Still stuck
Talk to a person
If you are not yet a customer, or you do not know which channel you need, start here and we will route it.