The problem
Nobody trusts an AI SRE they haven't seen work on their own stack.
AI SRE agents are being bought on demos and sales-led trials. Buyers only learn how one handles their telemetry after weeks of setup.
of AI SRE tool failures happen because the data they needed was never captured.
Lightrun, State of AI-Powered Engineering 2026of agentic AI projects will be cancelled by the end of 2027, over cost, unclear value and risk.
Gartner, June 2025of enterprises will use AI SRE tooling by 2029, up from under 5% in 2025.
Gartner Market Guide for AI SRE Tooling, Jan 2026How the game works
Get paged. Investigate. Name the cause.
Every puzzle comes from a real fault we inject into a running shop. The page, graphs and service health are free. Logs, traces and recent changes cost points. Wrong guesses cost more.
Get today's incident
One new outage a day. Aiden has already played, and its score is the one to beat.
Race Aiden, then share
Points for accuracy, time to break ties. Share your result in Slack or on LinkedIn.
See how Aiden SRE solved it
Every step of Aiden SRE's reasoning, including the trap it avoided.
Proving Ground
Would Aiden SRE solve this on your stack? Find out before you buy.
We copy your system into a throwaway environment, break it on purpose, and grade Aiden SRE against the cause we know. Production is never touched.
Your services in a throwaway Shipyard environment.
Inject a known fault. Real alerts fire in ObserveNow.
Aiden SRE works the alerts like a real incident.
A scorecard: which faults it solved, and what to fix.
What your Proving Ground scorecard looks like
| Injected fault | Generic AI agent | Aiden SRE |
|---|---|---|
| Payment failures from a feature flag | MISS Blamed checkout | PASS Named the flag change |
| Payment service unreachable | PARTIAL Found payment, no cause | PASS Dead dependency, from traces |
| Memory leak in recommendations | MISS Blamed high traffic | PASS Cache growth, from runtime metrics |
| Database lock contention in the catalog | PARTIAL Slow catalog, no cause | PARTIAL Slow queries, lock not named |
| CPU saturation in the ad service | PASS | PASS |
| Score | 1 / 5 | 4 / 5 |
How it grows, and how it pays
Every share is an ad. Every proof is a lead.
- Play
A free daily puzzle. Streaks. Two minutes.
- Share
Posted to LinkedIn and Slack.
- Prove
"Would Aiden SRE solve this on my stack?" A free Proving Ground run answers it.
- Buy
A paid Proving Ground plan: scorecards on your own stack, every time the agent changes.
Why it spreads. Every share names Aiden SRE. LinkedIn reports 84% of its game players come back the next day, and Wordle grew on the same share-a-grid loop. Nobody has built this for engineers.
LinkedIn, via Content Marketing Institute, April 2025
Competition
Nobody owns the daily habit.
Incident tools wait for an outage. AI SRE startups sell through demos. Game days happen once a quarter, with no AI and no grade.
| Can You Beat Aiden? with Aiden SRE | PagerDuty SRE Agent | incident.io Rootly | AI SRE startups Resolve, Cleric | Game days Gremlin, manual | |
|---|---|---|---|---|---|
| Daily habit | Yes | No | No | No | Quarterly |
| Shareable by design | Yes | No | No | No | No |
| Shows the AI's reasoning on a real incident | Every day | In Slack | Partly | In a demo | No AI |
| Proves the AI on your stack before you buy | Proving Ground | No | No | Sales-led POC | No grading |
Plans
Free to play. Paid to prove.
Play
Free- Daily puzzle and leaderboard
- One Proving Ground run on our demo shop
Team
$1,500 / month- 2 environments
- 25 drills a month
- Scorecards and fix lists
Enterprise
from $40k / year- Custom fault library
- A CI gate on every agent change
- Private runner, SSO
AI SRE vendors
from $60k / year- Benchmarks on real stacks
- "Proving Ground verified" badge
Early-access pricing, set with our design partners.
Questions
The tough ones, answered
Isn't a game a gimmick?
It's the free front door. Engineers meet Aiden SRE every day, long before a sales call. The business is Proving Ground.
How long does a Proving Ground run take?
About 2 minutes of your time: pick a fault and press start. Proving Ground breaks the copy, Aiden SRE investigates, and the scorecard arrives when it's done.
Where do the incidents come from?
Real faults injected into a running open-source shop on Shipyard, with real metrics, logs and traces.
Does it touch production?
Never. Proving Ground only breaks a throwaway copy or a test environment you choose.
Scan to play