> ## Documentation Index
> Fetch the complete documentation index at: https://docs.sureops.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Evidence-first AI SRE: BYOK, managed diagnosis & the sample-app trial

> How sureops earns trust before it acts — the evidence bundle, the BYOK vs. managed-diagnosis split, and the hands-on sample-app trial.

*This page is generated from `feature-catalog.yaml` — do not hand-edit; regenerate via the catalog generator and re-run `/update-public-docs`.*

sureops is evidence-first AI SRE: every diagnosis carries citations to the logs, traces, and deploys it reasoned from, a blast-radius view with a rollback plan for any recommended action, and a full decision trace you can audit before you trust it — on every plan, starting at Free.

## The evidence bundle — every diagnosis, every plan

Once a diagnosis exists, its full reasoning is yours to read — this is never a paywalled view, on any tier:

* **RCA with citations** to the logs, traces, and deploys the agent reasoned from
* **Blast-radius view** — what else this incident (or a proposed fix) could affect
* **Rollback plan** and risk explanation for every recommended action
* **Full decision trace** — alternatives considered, tool calls, evidence, all of it

**Free plan**: Connect and manage your own infrastructure as a sureops environment. Use your own LLM provider API keys for managed diagnosis and chat. No sureops-managed diagnosis runs on the Free plan and no managed spend is incurred — Free pairs your own telemetry with your own LLM key, and the full evidence bundle above renders for every diagnosis that key produces.

## Bring your own key, or let sureops manage it

|                        | Bring your own LLM key (BYOK)                  | sureops-managed diagnosis                                            |
| ---------------------- | ---------------------------------------------- | -------------------------------------------------------------------- |
| Who pays for the model | You — your provider account, your model choice | sureops — included in your plan's episode quota                      |
| Available from         | Free                                           | Starter                                                              |
| What you get           | The same evidence bundle above, on your key    | The same evidence bundle above, metered by diagnosed-episode credits |

**sureops-managed diagnosis is designed but not yet available for self-serve signup** — it requires a signed sureops enterprise agreement with the underlying LLM provider, which doesn't exist yet. Every organization today runs on BYOK (or a demo/trial environment); the columns above describe the pricing shape once managed mode launches. See [BYOK vs Managed LLM](/next/ai-agents/byok-vs-managed) for what's live today.

**Diagnosed-episode credits, not a raw spend meter.** A billable episode is a managed diagnosis that persists an evidence bundle — a correlated duplicate alert folded into an existing incident, a triage-only suppression, or a BYOK diagnosis never counts against your quota.

|                                             | Free | Starter | Growth | Enterprise        |
| ------------------------------------------- | ---- | ------- | ------ | ----------------- |
| Included managed-diagnosis episodes / month | 0    | 100     | 500    | Contract-governed |

Predictable episode quotas — beta: no billing. Full quota + overage detail lives on the [pricing & tiers page](/next/getting-started/pricing-and-tiers).

## Hands-on trial: the hosted sample app

Included from Starter: deploy a sureops-managed sample application stack and inject controlled failures to watch the full incident pipeline run end to end — no production access required. The trial is bounded by both a time window and a usage cap so it stays a trial: once either limit is reached, deploy and chaos-injection pause with an upgrade or convert prompt.

Ask in your onboarding channel (or see the [pricing & tiers page](/next/getting-started/pricing-and-tiers)) for current trial availability.

## What to expect from your value report

**Included from Starter.** Per-incident autonomous-vs-guided savings estimate — an honest estimate, methodology always stated. Here's the honest shape of it, stated up front so nothing it shows is a surprise:

* The only measured comparison is **your own autonomous-mode incidents against your own guided-mode incidents** — sureops has no measured manual baseline for your environment, so the report will never claim a "vs. manual" number.
* Guided-only accounts see **measured facts** — time-to-RCA, evidence assembled, your MTTR trend — never a fabricated counterfactual.
* A numeric estimate only appears once there's enough comparable incident history to support one; below that threshold, the report shows the evidence timeline with a matching confidence label instead of a headline number.