Microsoft shows how to build a personalized Voice Agent with enterprise knowledge
**Microsoft’s new Voice Agent walkthrough combines Foundry IQ knowledge, Personal Voice and a custom photo avatar in one real-time conversation.
** Once those assets are prepared, a supported Foundry project can connect them through the portal playground without writing speech, avatar or application code to try the experience.
Users can ask follow-up questions grounded in approved documents, while knowledge and presentation assets can be updated independently.
Personalization requires authorized documents and likenesses, recorded consent and applicable access approvals. Models, regions and preview features vary, and Personal Voice and photo avatars have eligibility requirements.
The article provides implementation guidance and example scenarios; it does not establish blanket general availability.
Partner briefing
- What it is
Microsoft published a Foundry Voice Agent implementation walkthrough.
- What’s changing
The walkthrough connects enterprise knowledge, a selected voice and an avatar in the portal playground.
- Why it matters
The agent can answer from approved documents and acknowledge gaps in its connected knowledge.
- Who it affects
Developers and partners building conversational product, support and training experiences.