Pre-release validation for energy platforms
Your tests are green. The charger never got the schedule.
We test the whole chain before every release.
From your platform to your partners' systems and the charger or battery at the end. Before every release, and every night, so you also catch what a partner changed in between.
Scan my 3 critical flows, free · 2 weeks · about 2 hours of your team's time · you keep the report · never on production
Your platform: sent. Partner's system: forwarded. Chargers: 1 in 6 got nothing. Your tests say all passed; breakmy.AI checks that not every charger got it. Without breakmy.AI: Monday 09:00, your customer calls. With breakmy.AI: caught before the release. Illustrative example.
Thomas Noë, founder. Nearly 20 years leading software quality at Collibra, Teamleader and Luminus, including in energy. Open to 5 energy platforms for a free scan in Q4 2026.
Four steps. No code changes.
- You name 3 flows — In plain words: “Every charger must receive its schedule.” The agent turns them into checks. You approve them.
- Before every release — The whole chain, start to finish, on your test environment. Ship or hold, in plain language.
- Every night — The same checks again, so a partner who changed something between your releases shows up the next morning.
- Real failures only — A failure is re-run and checked against the partner's own status before it turns red. Sandbox noise is flagged, not alarmed.
It starts with what your team already knows.
No integration project. For the free scan you tell the agent four things in plain language, in the setup call or in the chat. It turns them into checks you approve.
What you tell it: what must never fail, what went wrong before, how your partners respond, what changes in this release. Checks it proposes, you approve: every charger brand accepts the schedule, a slow partner like in March, the retry change in this release.
Nothing runs without your approval. Test environments only. EU-hosted, or use your own AI model key. On our roadmap: reading code changes and documentation directly.
Your tests check your software. We check the whole chain.
Your end-to-end tests check that your screens and your own flows work. Contract tests check that a partner's system accepts your request. breakmy.AI also checks that the charger, battery or partner actually did it, writes and keeps the checks up to date for you, and gives product a ship-or-hold report it can read.
When did a release last break something nobody saw? Who found it: your team, or your customer? What did finding the cause cost?
Your partners change things too. The nightly run catches it.
Every night the same checks run again on your test environment, so a change on a partner's side shows up the next morning, with the evidence. Not on a Saturday, from a customer.
Example nightly run: prices from the tariff partner arrive, pass. Chargers accept the new schedule, fail since 23:40. Batteries follow their schedule, pass. In plain language: a charger partner updated its test system last night. Since then, 1 in 6 chargers rejects your schedule. You didn't ship anything.
Three flows. Two weeks. A report your whole team can read.
- Week 1: a 15-minute call to pick the three flows, then we connect to your test environment.
- Week 2: the agent validates them on every release and every night.
- Readout, 45 minutes: what works, what broke, and what an incident would have cost you. A ship-or-hold report you can present internally.
No partner test system yet? We start with your own test environment and simulators. Free, no strings. If we find something, you decide whether the checks keep running.
Pick a 15-minute slot with Thomas or request your free scan with your work email, or write to hello@breakmy.ai.
What your team will actually deal with.
- What we need — Access to your test environment and test accounts at your partners. API credentials scoped by you. Nothing to install in your product.
- Where results go — Into your release pipeline as ship or hold, plus Slack, Teams or email. Sandbox noise is reported separately, so red means a real failure.
- Checks you can read — Every check is plain language plus code you can review, export and keep in git. Nothing changes without your approval.
- When a partner changes — The agent proposes an updated check and shows the diff. A broken contract stays red until you decide.
Your data and your assets stay yours.
- Never on production — We validate on your test environment and simulators. Never on live chargers or batteries.
- Your credentials, scoped by you — Read access where possible, stored encrypted in the EU, revocable in one click.
- EU-hosted, no training — Data stays in the EU. The AI provider does not train on your data.
- Every action logged — An audit log of every check and change. Security one-pager and DPA before you start.
Start free. Then one plan per platform.
Priced per platform, not per seat. No commitment after the scan.
- Team — €299 / month — One platform, core flows
- Scale — €999 / month — Many assets, vendors or products
- Enterprise — On quote — Self-hosted runner, SLA, audit trail
Questions we get
We already have tests. Why this? Your tests check your own code and screens. We check that the whole chain works: your platform, your partners' systems and the charger or battery at the end.
What about failures in production? We validate before you go live and every night on your test environment. Watching production is your monitoring tool's job; we make sure less reaches it.
Our partners' test systems are flaky. We know. Failures are re-run and checked against the partner's own status before they turn red, and sandbox noise is reported separately.
What if the founder is unavailable? Checks run on the platform and results go straight into your release pipeline and your team's channels. You get the checks as code.
What does the scan cost? Nothing, and no commitment. About two hours of your team's time.
Which three flows must never fail?
Tell us, and we validate them for free for two weeks. Scan my 3 critical flows, free
breakmy.AI · Belgium · For engineers · EU-hosted · Privacy · Log in