Deployment
Promptev runs your agents on Promptev cloud with your own AI model keys, and lets you keep files in your own storage bucket and the search index in your own database, so your data stays in accounts you control.
Your cloud
Your AI modelon your own keys
Promptev enginereads, searches, answers
Your storageyour own bucket
Your databasethe search index
Your teamgets the answer, nothing else
How it works
Step 1
Add your own API keys for OpenAI, Anthropic, Google Gemini or DeepSeek and choose a model for each agent. Your agents’ model calls go to your provider on your account, so its terms and your billing apply.
Switch AI models without rebuilding a single agent.
Then decide where your files live.
Step 2
Files live in managed storage by default. Connect your own Amazon S3 (or S3-compatible), Google Cloud Storage or Azure Blob Storage bucket instead, and every file moves with it.
Your files sit in a bucket you own.
The search index can be yours too.
Step 3
Promptev manages the index of your documents for you, or you point it at your own PostgreSQL database with your own embedding model. The knowledge graph is stored in that same database, and you can add your own reranker if you want one.
What Promptev learns from your documents stays in your database.
Where your data lives
Need it inside your own environment? Talk to us.
Example
A law firm runs its contract agents on its own Anthropic account, so model usage appears on its own bill under its own terms. Matter files sit in the firm’s Azure storage, and the search index in its own PostgreSQL database.
When the firm later tries a different model for one agent, it changes one setting. The agent, its knowledge and its approvals stay as they were.
Questions
Models from OpenAI, Anthropic, Google (Gemini) and DeepSeek, on your own API keys. Each agent can use a different model, and you can change it later without rebuilding the agent.
On Promptev cloud, which runs on Google Cloud in the United States (us-east4). Files can live in your own Amazon S3, Google Cloud Storage or Azure Blob Storage bucket instead, and the search index in your own PostgreSQL database.
No. Promptev does not use your content to train its own models or anyone else’s without your explicit opt-in. Model calls made on your keys follow your agreement with your AI provider.
You choose: 30, 60, 180 or 365 days, or forever if the files are in your own bucket. Your knowledge documents are not affected by this setting.
Promptev cloud is the standard way to run it, with your models, storage and database connected as above. If you need Promptev inside your own environment, talk to us.