customizing_azd_parameters.md

Back to Chat with your data README

Overview

By default the deployment uses your environment name as a prefix to keep Azure resource names unique. The parameters below show their default values; you only need to set a parameter when you want to change it.

azd up prompts you only for values that are unset and have no default, such as the environment name and the deployment region. Every other parameter listed below ships with a safe default (shown in the tables) and is used unless you override it. To override any parameter, run azd env set <NAME> <VALUE> before azd up. On the first azd command you are prompted for the environment name; choose a 3–16 character alphanumeric name.

Core configuration

Name	Type	Default	Purpose
AZURE_ENV_NAME	string	(prompted)	Environment name prefix for all resources (3–16 alphanumeric characters).
AZURE_LOCATION	string	(prompted)	Region for the resource group and regional resources.

Database

The database type is chosen once and is locked after deployment. It sets both the chat-history store and the retrieval index.

Name	Type	Default	Purpose
DATABASE_TYPE	string	postgresql	postgresql (PostgreSQL Flexible Server + pgvector) or cosmosdb (Cosmos DB + Azure AI Search).

See Chat history and PostgreSQL.

Azure AI Foundry models

Name	Type	Default	Purpose
AZURE_AI_SERVICE_LOCATION	string	(AZURE_LOCATION)	Region for Azure AI Services and Foundry; restricted to regions with capacity for the chat model.
AZURE_GPT_MODEL_NAME	string	gpt-5.4-mini	Chat model name.
AZURE_GPT_MODEL_VERSION	string	2026-03-17	Chat model version.
AZURE_GPT_MODEL_DEPLOYMENT_TYPE	string	GlobalStandard	Chat model deployment type.
AZURE_GPT_MODEL_CAPACITY	integer	50	Chat model capacity (TPM, thousands).
AZURE_REASONING_MODEL_NAME	string	gpt-5-mini	Reasoning model name.
AZURE_REASONING_MODEL_VERSION	string	2025-08-07	Reasoning model version.
AZURE_REASONING_MODEL_DEPLOYMENT_TYPE	string	GlobalStandard	Reasoning model deployment type.
AZURE_REASONING_MODEL_CAPACITY	integer	50	Reasoning model capacity (TPM, thousands).
AZURE_EMBEDDING_MODEL_NAME	string	text-embedding-3-small	Embedding model name.
AZURE_EMBEDDING_MODEL_VERSION	string	1	Embedding model version.
AZURE_EMBEDDING_MODEL_DEPLOYMENT_TYPE	string	Standard	Embedding model deployment type.
AZURE_EMBEDDING_MODEL_CAPACITY	integer	100	Embedding model capacity (TPM, thousands).
AZURE_OPENAI_API_VERSION	string	2025-01-01-preview	API version for chat and embedding calls.
AZURE_AI_AGENT_API_VERSION	string	2025-05-01	API version for the Foundry agent runtime.

See Model configuration and Model quota settings.

Ingestion

Name	Type	Default	Purpose
INGESTION_TRIGGER	string	direct_enqueue	How ingestion starts: direct_enqueue (admin uploads enqueue work) or event_grid (blob events trigger ingestion).

See Document ingestion.

Reliability and security

These flags align the deployment with the Well-Architected Framework and apply to the WAF deployment profile (main.waf.parameters.json).

Name	Type	Default	Purpose
ENABLE_TELEMETRY	boolean	true	Send anonymized module usage telemetry.
ENABLE_MONITORING	boolean	false	Deploy Log Analytics and Application Insights and wire diagnostic settings.
ENABLE_SCALABILITY	boolean	true	Higher SKUs and autoscale on Container Apps, Azure AI Search, and PostgreSQL.
ENABLE_REDUNDANCY	boolean	false	Zone-redundant and paired-region failover on the data and compute resources.
ENABLE_PRIVATE_NETWORKING	boolean	true	Deploy a virtual network, private endpoints, and Bastion, and disable public network access on data-plane resources.
AZURE_ENV_VM_ADMIN_USERNAME	string	(empty)	Jumpbox VM admin username (used when private networking is enabled).
AZURE_ENV_VM_ADMIN_PASSWORD	string	(empty)	Jumpbox VM admin password (used when private networking is enabled).
AZURE_ENV_JUMPBOX_SIZE	string	Standard_D2s_v5	Jumpbox VM size (used when private networking is enabled).

Bring your own resources

To reuse existing resources instead of provisioning new ones, set the matching resource ID. Leave a value empty to have the deployment create the resource.

Name	Type	Default	Purpose
AZURE_ENV_LOG_ANALYTICS_WORKSPACE_ID	string	(empty)	Resource ID of an existing Log Analytics workspace.
AZURE_EXISTING_AIPROJECT_RESOURCE_ID	string	(empty)	Resource ID of an existing AI Foundry project.

How to set a parameter

Set a parameter before azd up:

azd env set <NAME> <VALUE>

Examples

Set the deployment region:

azd env set AZURE_LOCATION eastus2

Deploy in PostgreSQL mode:

azd env set DATABASE_TYPE postgresql

Turn on monitoring and private networking:

azd env set ENABLE_MONITORING true
azd env set ENABLE_PRIVATE_NETWORKING true

Notes

• Region availability. Not every service or model is available in every region. Confirm availability for your chosen region before deploying; see Check quota by region.
• Locked after deploy. The database type cannot change after deployment. To switch, deploy a new environment.
• Orchestrator selection. The chat orchestrator has no dedicated azd parameter. azd up sets it automatically from the databaseType choice: postgresql selects langgraph, and cosmosdb selects agent_framework. To run a different orchestrator than that default, switch it at runtime from the admin Configuration page with no redeploy; both orchestrators are served on either store. See Admin and configuration and Architecture overview.