Celeris Home
Test Execution
AI-first debugging

Test execution, built into the platform.

Run E2E, load, smoke, and synthetics across preview environments and production — then use results to automatically gate canaries.

Run your first test
Private runners (BYOC)
Canary gates
Artifacts + traces
AI debugging
Test Service Result
Live

Video placeholder

60-second test execution overview
How It Works

Test execution as a first-class platform primitive

From environment selection to AI-powered debugging — every step of the test lifecycle, orchestrated.

Environment: staging Gateway Service A Service B Private Service C Runner Managed Canary Gate RESULTS 24/24 Passed p95: 142ms ✨

Telemetry

Idle
Environment staging
Test Type Smoke
Runner Managed

Metrics

Latency p95
—
Error Rate
—
Throughput
—
Pass Rate
—
p95 < 200ms —
Error < 0.1% —

Artifacts

junit-results.xml
screenshots/
k6-summary.json

✨ AI Findings

All tests passed. Service B response time improved by 12% compared to the baseline. View trace correlation →

Test Types & Runners

Every test type, every execution mode

From quick smoke tests to heavy load simulations — run them on managed runners or bring your own.

Smoke

Fast health checks

<30s

E2E Browser

Playwright/Cypress

Screenshots

API

Postman/Newman

REST/gRPC

Load/Perf

k6 thresholds

p95/RPS

Synthetic

Multi-region probes

Scheduled

Contract

Pact-style

Provider

Managed Runners

Zero config, instant scale

  • Auto-scaling runner pools (0→∞)
  • Browser environments (Chrome, Firefox, Safari)
  • Pre-warmed for 5s cold starts
  • Multi-region (US, EU, APAC, LatAm)
Public endpoints only Shared infra

Private Runners

BYOC

Your VPC, your data plane

  • Runs inside your VPC/BYOC cluster
  • Access private services via internal DNS
  • GPU runner pools for AI/ML test loads
  • Network policy controls (service allow-lists)
Private services Dedicated infra

Trigger tests from anywhere

Manual (UI/CLI)
PR/Preview auto
Scheduled/Cron
Release gates
Webhook/API
Private Service Testing

Test private services without public ingress

Private runners inside your environment reach internal services directly via service discovery — no exposure needed.

ENVIRONMENT: STAGING PRIVATE User traffic Gateway public Service A api-gateway Service B payment-processor PRIVATE Private Runner in your VPC DIRECT TEST PATH Internal DNS / east-west No public ingress Legend: Test probe (direct) User traffic (bypassed)

Test probe reaches Service B directly via internal routing

Zero-trust compatible

Tests authenticate via workload identity; no static credentials.

Service discovery

Resolve services by name; no hardcoded IPs or external DNS.

Network telemetry

Full trace visibility even for internal-only test flows.

Canary Gates

Tests that stop bad deploys

Configure tests as progressive delivery gates. Canaries run tests at each step — fail means automatic stop and rollback.

Deploy Started
5% Traffic Smoke ✓
25% Traffic E2E Running...
3
50% Traffic Load Test
100% Promoted

Gate 2: E2E Tests

In Progress
checkout-flow.spec.ts Passed
user-auth.spec.ts Passed
payment-integration.spec.ts Running
inventory-sync.spec.ts Queued
Progress 2/4 tests

Canary Metrics (25% traffic)

Latency p95
142ms
Threshold: <200ms ✓
Error Rate
0.02%
Threshold: <0.1% ✓
Success Rate
99.98%
Baseline: 99.95%
Traffic Split
25%
Next: 50%

What happens on failure?

If tests fail or metrics breach thresholds, the canary automatically stops traffic shift and initiates rollback to the last stable version. You get alerts + AI root cause analysis.

Auto-rollback enabled

Pre-canary gate

TestsSmoke suite
Timeout5 min
RequiredYes

Mid-canary gate

TestsE2E + Load
Thresholdsp95 < 200ms
RollbackAuto

Post-canary gate

TestsFull regression
ApprovalManual
Bake time30 min
Results & Visualization

Results you can actually act on

Structured results, artifacts, trend charts, and deep trace linking — everything you need to understand what happened.

Test Run #2847

Passed
staging / PR-128 • 2m 34s ago
24
Passed
0
Failed
0
Skipped
2:34
Duration
Test Suites
checkout-flow.spec.ts
8 tests 0.8s
user-authentication.spec.ts
6 tests 1.2s
payment-integration.spec.ts
5 tests 0.4s
inventory-sync.spec.ts
5 tests 0.2s
Test Duration Trend (last 20 runs) Avg: 2m 28s

Artifacts

junit-results.xml
12 KB
screenshots/
24 files
video-recording.webm
8.4 MB
k6-summary.json
4 KB

Observability

View traces View logs View metrics

Metadata

Commit a1b2c3d
Branch feature/checkout-v2
Runner managed-us-east
Triggered by PR merge

Outputs in standard formats for any integration

JUnit XML JSON Webhooks S3 export REST API
✨ AI-Powered

Your AI test engineer

Generate tests, debug failures, and ship fixes with guardrails. The AI agent correlates test results with traces, logs, and recent deploys.

Generate Tests

From your API specs

AI Agent

Generate E2E tests for my checkout API from the OpenAPI spec

I've analyzed your checkout-api.yaml and generated 12 test cases covering:

  • Cart creation and modification
  • Payment processing (success/failure)
  • Edge cases (empty cart, invalid coupon)
OpenAPI / Swagger gRPC reflection GraphQL schema

Debug Failures

Root cause analysis

payment-integration.spec.ts:42

Expected status 200, received 504 (Gateway Timeout)

✨ AI Root Cause Analysis

The failure correlates with a spike in payment-service latency starting 3 minutes ago, following deploy v2.1.4.

View correlated trace →
View deploy diff →
Trace correlation Log analysis Deploy timeline

Suggest thresholds

Learns optimal k6 thresholds from your historical SLOs

Auto-fix suggestions

Proposes config changes, timeouts, retry logic

Flaky test detection

Identifies unstable tests and suggests stabilization

PR diff checks

Suggests tests based on code changes in PRs

Integrations

Tests with ephemeral dependencies

Hook into the Integration Marketplace to provision ephemeral databases, queues, and feature flags — tests run against isolated, seeded environments.

Test Run Starts
provisions
Ephemeral DB Seeded + isolated
runs against
Services In slate

Databases

PostgreSQL, MySQL, MongoDB

Branch databases or seed from fixtures. Automatic cleanup after test run.

Neon PlanetScale Atlas

Queues & Streams

Kafka, SQS, Redis

Isolated topics/queues with pre-loaded messages for integration tests.

Upstash Confluent

Object Storage

S3-compatible buckets

Ephemeral buckets with test fixtures. Perfect for upload/download tests.

Cloudflare R2 MinIO

Feature Flags

Test flag variations

Override flags per test run to validate feature behavior in isolation.

LaunchDarkly Split

Secrets

Test credentials

Inject test-specific secrets securely. Auto-rotated after test completion.

Vault Doppler

CI/CD Hooks

GitHub, GitLab, Jenkins

Trigger tests from your existing pipelines with full status reporting back.

GitHub Actions GitLab CI
Explore Integration Marketplace
Tests running everywhere

Ship with confidence,
not with fingers crossed.

Test execution built into the platform. Private runners, canary gates, AI debugging, and deep observability integration — all in one place.

Run your first test — free Talk to an expert

Trusted by engineering teams at

SOC 2 Type II
BYOC/Private runners
99.99% SLA
Enterprise support