Authava and Microsoft Azure
If your business runs on Microsoft Azure, your bot fits right into it. By default, the bot's AI runs on OpenAI's GPT models hosted on Azure. During a conversation, the bot can call your Azure Functions and APIs. Afterward, conversation data can flow into your own Azure storage.
Why Would An Azure Shop Choose Authava?
You've already chosen Azure for your apps, your data, and your security reviews. A chatbot shouldn't pull your customer conversations out into a separate world. With Authava:
- The AI runs on Azure. Your security team reviews a provider it already knows.
- Your Azure services do the work. The bot calls the Functions, Logic Apps, and APIs you've already built, instead of you rebuilding that logic in a chatbot.
- Your data lands in your Azure. Conversations can be stored where the rest of your data lives, for your own reporting and AI work.
Authava handles the part Azure doesn't: the customer conversation itself, answered from your approved content, with handoff to people and reporting on what customers ask.
Does Authava Run On Azure?
By default, Authava's bots answer using OpenAI's GPT models hosted on Microsoft Azure. Our AI providers' business terms state that data sent through their APIs isn't used to train their models, and Authava doesn't use your conversations to train public AI models.
Authava itself runs primarily in the United States, on providers including Microsoft Azure, AWS, and DigitalOcean. See Data Privacy for the full detail.
Can The Bot Call My Azure Services?
Yes. Authava's Full Integration calls your web endpoints at each stage of a conversation, and Azure Functions, Logic Apps, and APIs behind Azure API Management are all web endpoints. For example:
- When a conversation starts, an Azure Function can look up the visitor and shape the bot's first message.
- During the conversation, the bot can call your Azure-hosted APIs as tools through the MCP Gateway, for live answers like an order status or an account balance.
- Once a conversation appears finished, Authava can send it to your Azure Function or Logic App to route the lead or start a workflow.
Can Conversation Data Go Into My Azure Storage?
Yes. Authava has an API, and it gives you two ways to get your data into Azure:
- Pull it. Your systems call Authava's API to pull transcripts into Blob Storage, Azure SQL, or your data platform, on whatever schedule you like.
- Have Authava push it. Authava can send each finished conversation to an endpoint you choose, such as an Azure Function or Logic App.
Our team can connect it as part of your setup, or your own team can. Either way, your storage stays under your control.
That gives you your customer conversations next to the rest of your data, for your own dashboards, analysis, or AI projects.
Can My Azure Systems Keep The Bot Up To Date?
Yes. Your systems can call Authava's API to upload updated documents, so the bot answers from current information without anyone uploading files by hand.
Does It Work With The Rest Of Microsoft?
- Calendars. The Scheduling module books onto Microsoft 365 calendars as well as Google Calendar.
- Sign-in. For a Portal Bot, Authava is built to work with the sign-in system you already use, so signed-in users don't log in twice. See Authava and AWS Cognito for how that works.
Frequently Asked Questions
Does Authava work with Microsoft Azure?
Yes. By default its AI runs on GPT models hosted on Azure, the bot can call your Azure Functions and APIs, and conversation data can be sent to your Azure storage.
Can Authava send our chat transcripts to Azure Storage?
Yes. Your systems can pull transcripts through Authava's API, or Authava can push each finished conversation to your Azure Function or Logic App.
Can the bot call our Azure Functions during a conversation?
Yes. Through Full Integration and the MCP Gateway, the bot can call your Azure Functions and APIs and use the result in its reply.
Is our data used to train AI models?
No. Authava doesn't use your data to train public AI models, and our AI providers' business terms state that API data isn't used to train their models.