Everything you need to test on real infrastructure
A full walkthrough of what Society Labs does, how it fits your stack, and why real EC2 beats containers for this job.
Built for teams who test on real infrastructure
Everything you need to catch environment-specific bugs before your users do.
Real EC2 environments
Every test run happens inside an actual, freshly provisioned EC2 instance — the same instance types you deploy to in production.
Fast, disposable runners
Environments boot in under two minutes and are torn down automatically when your suite finishes. No lingering costs, no cleanup.
Matrix testing across instance types
Run the same suite across t3, m5, c6i, and GPU-backed instances in parallel to catch instance-specific failures.
Deep infrastructure telemetry
CPU steal time, network jitter, disk I/O — see the metrics that only show up on real hardware, correlated with test failures.
Plug into your CI
Drop-in integrations for GitHub Actions, GitLab CI, and Jenkins. Trigger EC2-backed runs from your existing pipeline.
Isolated by default
Every run gets its own VPC and security group. Nothing persists, nothing leaks between test runs.
Historical run comparison
Every run is versioned automatically, so you can diff infra metrics and results against last week's baseline in one click.
Spot & chaos injection
Opt into simulated spot-interruption and network-partition events mid-run to see how your app actually degrades.
Snapshot & replay
Freeze a failing environment exactly as it was and replay it locally or share it with a teammate for debugging.
How it works
Connect your repo
Point Society Labs at your test suite and tell us which instance types matter to you.
We provision real EC2 instances
On every run, we spin up fresh, isolated EC2 environments matching your target configuration.
Your tests run against real infrastructure
No mocks, no containers pretending to be VMs — your suite executes on the real thing.
Get results plus infra-level insight
See pass/fail alongside the infrastructure metrics that explain the flaky and environment-specific failures.
Wherever your bugs hide, this catches them
Different teams test on EC2 for different reasons. Here's where Society Labs fits.
Catch multi-tenant load issues before rollout
Run your full regression suite against m5 and c6i instances under simulated concurrent load, so noisy-neighbor and autoscaling bugs surface in CI, not in an incident channel.
Validate against production-shaped network topology
Reproduce your exact VPC, subnet, and security group layout for every test run, so latency-sensitive transaction paths get tested the way they'll actually run.
Stress-test checkout before a launch or sale
Spin up a fleet of EC2 test runners to hammer your checkout flow at real traffic shapes, weeks before Black Friday instead of during it.
Test on the GPU instances you actually deploy to
Run inference and training smoke tests on g4dn/g5 instances so CUDA driver mismatches and memory ceiling issues show up before a model ships.
Kill "works on my machine" for good
Every engineer's test run happens on the same disposable EC2 baseline, not their laptop or a shared staging box that's slowly drifted from prod.
Catch jitter and packet loss under real network conditions
Test latency-sensitive multiplayer and streaming paths against real EC2 network characteristics instead of a loopback interface on a container.
Why not just use containers or a manual EC2 box?
Containers are fast but they lie about the environment. Hand-rolled EC2 scripts are accurate but slow to maintain. Society Labs is built to be both.
| Docker / containers | DIY EC2 scripts | Society Labs | |
|---|---|---|---|
| Runs on real EC2 hardware | ✕ | ✓ | ✓ |
| Matches production instance type | ✕ | Manual | ✓ Automatic |
| Spin-up time | ~seconds | Minutes–hours | ~90 seconds |
| Automatic teardown | ✓ | Manual / scripted | ✓ |
| Infra-level telemetry (steal time, I/O, jitter) | ✕ | DIY instrumentation | ✓ Built in |
| Multi instance-type matrix runs | ✕ | Custom tooling | ✓ Native |
| Chaos / spot-interruption injection | ✕ | Custom scripting | ✓ One click |
| CI integration | ✓ | Custom scripts | ✓ Drop-in |
Fits into the pipeline you already have
No migration required — trigger EC2-backed runs from your existing CI, and pipe results to the tools your team already watches.
Isolated, ephemeral, and boring by design
Security in a testing platform should be uneventful. Here's exactly what that means.
- Fresh VPC per run — every test run gets its own isolated network, torn down immediately after.
- No persistent storage — instances are ephemeral by default; nothing from one run is visible to another.
- Least-privilege IAM — scoped roles generated per run, not a shared long-lived credential.
- Your AWS account or ours — run inside your own AWS account for full control, or ours for zero setup.
- Full audit trail — every provision, run, and teardown is logged and exportable.
Ready to see pricing?
Founding-member rates are locked in for waitlist signups during the beta.