Integrations — Model providers
NerveStax and Azure OpenAI
Your Azure endpoint, your deployment, your key and your agreement.
What we do
What NerveStax does with Azure OpenAI
If your models already run in Azure under a procurement your security team has signed off, point NerveStax at them. You give the endpoint, the API version, the deployment name and the key; agents then run entirely within that arrangement. There is no fallback key behind the product, and calls go from your deployment to Azure, never through us.
-
Endpoint, version and deployment
Configure the Azure endpoint, the API version and the deployment name alongside the key. Agents address your deployment by name, so region and network rules stay under Azure's control.
-
Your key, per organization
Each organization holds its own key. NerveStax ships no fallback key, so with none configured an agent stops with a clear error rather than calling out on another account.
-
Choose the model per agent
Point different agents at different deployments: a small one for context lookups and alert triage, a larger one where a change is being designed and reviewed.
-
Calls are never proxied by us
Requests go from your NerveStax deployment to your Azure endpoint. The product ships no third-party analytics or error-tracking SDKs to carry anything elsewhere.
-
Spend tracked per call
Every call is recorded with organization, workspace, user, conversation, agent, model and token counts, so cost is attributable to a team and an agent rather than one Azure line item.
Background
About Azure OpenAI
Azure OpenAI serves OpenAI models inside a Microsoft Azure subscription, with the region, network controls and commercial terms of the rest of that tenancy. Models are addressed as deployments.
How it connects
What connecting it looks like
-
01
Deploy a model in Azure
Create the Azure OpenAI resource and deploy the model you want, then note the endpoint, deployment name and API version.
-
02
Add the provider in settings
Enter the key with the endpoint, API version and deployment. The key is encrypted at rest and no screen shows it again.
-
03
Assign deployments to agents
Map each agent to a deployment. Start smaller for context and triage work and raise it where the reasoning falls short.
-
04
Check usage after a week
Read the token usage page by agent and workspace, and rebalance the deployments from what it shows.
Scope
What this does not cover
- What reaches Azure: schemas and table metadata, dbt artifacts, repository files, command output and Airflow logs with secrets redacted, alert payloads, and, when an agent queries or samples a table, the rows it retrieves.
- Secrets are redacted from what the model reads. Result rows are not masked, so treat your Azure resource as in scope for whatever an agent is allowed to query.
- Region, retention and network controls for the model itself are Azure's, under your subscription and your agreement with Microsoft.
Questions
Azure OpenAI and NerveStax
Q01Why pick Azure OpenAI over OpenAI directly?+
Usually because the commercial and security review is already done: the models sit in a subscription your organization governs, in a region you chose, under terms already signed. NerveStax treats that as the provider and runs everything on it, rather than asking you to open a second account.
Q02What do we need to configure?+
The API key, the Azure endpoint, the API version and the deployment name. Those four are stored as the provider's configuration; the key is encrypted at rest with a key held outside the database, and no endpoint reveals it.
Q03What exactly is sent to Azure?+
Schemas and table metadata, dbt manifest, lineage and run results, repository file contents, command output and Airflow logs with secrets redacted, uploaded documents, alert payloads, and the rows an agent retrieves when it queries or samples a table. It goes to your resource under your key.
Q04Does NerveStax hold a key of its own?+
No. There is no platform fallback key of any kind. With no provider configured the agent raises a configuration error and does nothing, which means no work can run outside your own agreement with a provider.
Q05Can we keep prompts inside our own network entirely?+
Azure keeps them inside your Azure tenancy. If they must stay inside your own network, point NerveStax at an OpenAI-compatible endpoint you host and self-host the product itself; that combination keeps prompts within your boundary.