Your data, under your control
What is stored, where it is processed and with which AI model, so your security team can assess it on the facts.
Your data is queried, not copied
On every task, the agent looks up what it needs at that moment and doesn’t copy your databases. The code you provide for analysis is read in a temporary workspace, and a ZIP is kept for 8 days at most. What stays is the knowledge (map, catalog, documents, summaries and short code citations) and the record of every run.
Nothing is used to train models
What the agent reads is used for that task only. No AI provider trains its models on it.
You choose the model
The ones Reelease offers, inside the European Union, or a private model in your own infrastructure.
Europe by default, your region if you need it
Reelease services run in Madrid, the database in Frankfurt and the AI models inside the European Union: EU data residency for your systems’ information. With a private model or a deployment in your cloud or on your premises, everything stays where you choose.
Where your data is processed
- Reelease services
- Analysis, planning and connections run on Google Cloud in Madrid, and configuration, knowledge and run records are stored encrypted in Frankfurt. If your contract includes it, everything runs in your own cloud or on your premises.
- Your systems
- They connect through the API you publish or a Client Bridge installed in your network. Your databases are never exposed.
- AI models
- Inside the European Union with the models Reelease offers, or wherever you host your private model. Code analysis can run on open models on dedicated machines, without sending code to any external provider.
See also the privacy policy and the legal information.
The AI model, your choice
The models Reelease offers, inside the European Union, or your own private model in your infrastructure.
- Offered by ReeleaseDefault
- Your private modelYour choice
How it’s chosen
- For each system
- You choose the model its agents work with.
- For each agent
- You can change it for a specific agent; your choice for that agent wins.
- It never switches providers
- If the chosen model isn’t available, the task stops instead of moving to another provider.
- Inside the European Union
- The models Reelease offers run inside the European Union. Yours runs wherever you host it.
More detail in our subprocessor list.
Frequently asked questions
What data does Reelease keep?
Only what it needs to operate and audit: the configuration, the knowledge your team reviews (map, catalog, documents, per-file summaries and short code citations) and the record of every run. Everything is stored encrypted, with retention periods per system and legal hold.
Which AI models can we use?
The ones Reelease offers, inside the European Union, or a private model. You decide for each system and can adjust it per agent; if the chosen model isn’t available, the task stops instead of switching to another provider. No provider trains on your data.
What happens to the code I provide for analysis?
It’s used to understand your system. A repository is cloned into a temporary workspace that is deleted when the analysis finishes. A ZIP, up to 100 MB on Free, 1 GB on Pilot and 10 GB on Pro and Enterprise, is kept for 8 days at most, just long enough to finish the analysis, or less if you remove the source or the agent. What the model reads reaches the AI provider you use. The extracted knowledge, with summaries and short code citations, stays until you delete the agent.