Research data stays in the account it already lives in.
Exalt Reserve
For research organizations whose study and lab data already sits in AWS, Azure or Google Cloud. Exalt Reserve runs open models, retrieval and agents inside that same account, reachable only from your private network.
Where research starts.
Research already runs on commercial cloud
Through the NIH STRIDES Initiative, NIH-funded researchers get access to Google Cloud, AWS and Microsoft Azure, under partnerships that began in 2018.NIH STRIDES Initiative
Your account, your data
If your research data already sits in your own cloud account, Reserve deploys into that account rather than adding an Exalt-hosted service.
Discovery and regulatory use are treated apart
FDA's January 2025 draft guidance on AI to support regulatory decisions for drugs and biological products proposes a risk-based credibility assessment for each model's context of use, and leaves drug discovery out of scope.FDA's January 2025 draft guidance on AI for drug and biological products
Your review
Your organization runs its own review and stays responsible for its conclusions.
Same question. Different access.
Two people ask Exalt Reserve the same question. Each answer comes only from the documents that person can read.
Signed in as a study team scientist
- A study team scientist asks: "What did the interim results of the fibrosis study show?"
- The answer comes from the study report and cites it. The question, who asked and the source shown are recorded in the audit log.
- The citation opens the study report at "Interim results", with the passage highlighted.
Signed in as a lab operations coordinator
- A lab operations coordinator asks the same question.
- They cannot read the study report, so the answer never mentions it. It comes from the research data policy, which says who to ask.
- The citation opens the research data policy at "Unpublished results", and the audit log records the question, who asked and the one source shown.
Illustrative. A demonstration workspace, not a customer session.
What sits in your account.
- Models, retrieval, agents and monitoring run inside your own AWS, Azure or Google Cloud account. This is not an Exalt-hosted shared service.
- Open models self-hosted in your account are the default: Llama, Mistral, Qwen, Gemma and DeepSeek. Which models run, and whether a use case also uses an outside AI service, is decided with you.
- Reachable only from your private network, with no public endpoint. People sign in through your own identity provider.
- Data is encrypted at rest and in transit.
- Exalt runs a management plane outside your account. It holds no customer content.
- Two bills: Exalt's software, and your own cloud bill. You keep the data, models, vector stores, credentials and subscription.
- An audit log of questions, answers and admin actions is kept in your own cloud account, and your admin sets how long history is kept.
Start with the account you already run.
Bring the cloud account, the material you would put to work and the person who reviews your vendors. The consultation starts with those three.