RL environments

See the proof before you buy it.

Environment preview
Deterministic 2D environment
2D run
Incident bridgeSummaries, owners, and next actionsWar roomPagerStatusTimelineIncidentCoordinationSummariesAPI latencySEV-1Owner assignedSREStatus updatePosted

This is a working walkthrough of a shipped surface, not a mockup. A policy is submitted, gated before it can score, ranked, and settled to a creator payout. State, owners, and evidence travel with every step.

Read-only here. The numbers are illustrative seed data; in a pilot they read from your live metrics.

Agent loop

Every run earns a reviewer verdict, not just a score.

1ObservationAPI latency spike, stale channel summary, missing owner
2ActionAssign SRE, post update, request rollback approval
3Reward+0.91 for safe escalation and complete handoff
4Trace5 tool calls, 2 policy checks, 1 reviewer note
5ReviewApproved after status-page wording fix

Run checklist

No run touches the leaderboard until it clears the gate.

Sandboxpassed
Riskmedium
Ownerplatform ops
Runqueued
Leaderboard + payout handoff

Approved scoring updates the ranked submissions and creator settlement state together.

From the platform

The catalog, the leaderboard, the payouts. Real screens.

Catalog preview

20 environments ready

Leaderboard preview

Ranked after submission

Five illustrative leaderboard rows for the Slack incident-response environment
RankModelScore
#1
incident-runner-v4
p95 response quality
0.9692
#2
ops-policy-guard
approval-safe actions
0.9432
#3
slack-sre-copilot
handoff accuracy
0.9017
#4
oncall-summarizer
summary precision
0.8734
#5
triage-baseline
completion rate
0.8205

Leaderboard entries appear after submission and inherit the approval state of the environment.

Creator-dashboard preview

Payout state is visible

Payable$126.4
Accruing$38.2
Processed$248.9

Amounts accrue to $50 before payout.

The full loop

Catalog to creator payout, in five steps.

Every step carries its own evidence: who submitted, what passed review, and how it scored. In a pilot the same steps run on your policies.

  1. Catalog
    Step 01 · Catalog

    Browse seed environments

    Operators start with the same 20-item seed catalog used by the product surface.

  2. Detail
    Step 02 · Detail

    Inspect one environment

    The detail view keeps the preview, scoring history, and leaderboard in one place.

  3. Submit
    Step 03 · Submit

    Submit a policy with a budget cap

    The queued-run flow uses an idempotency key so a double-submit cannot create duplicate runs.

  4. Deployment status
    Step 04 · Deployment status

    Watch the queued run progress

    Queued, running, and completed states are visible before leaderboard updates are accepted.

  5. Creator dashboard
    Step 05 · Creator dashboard

    Review creator payout state

    Payable, accruing, and processed totals use the same seed values as the platform dashboard.

Revenue share

Build an environment, keep the larger share.

Base share, verified bonus, scale bonus, and payout threshold are the same constants the platform uses to settle creators. The math you see here is the math you get paid on.

70%
Base

creator share on approved community environments

+5%
Verified

bonus when the submission passes verified-environment checks

+5%
Scale

10,000 monthly downloads earn the scale bonus

$50
Minimum

settlements accrue until this payout threshold is met

RL Environments Demo | AuraOne | AuraOne