Microsoft 365 Copilot APIs: Unlocking enterprise knowledge for AI with the Retrieval API

Introduction

The Retrieval API, now available in Public Preview, provides developers with a secure, compliant, and scalable way to integrate enterprise content into AI workflows. Early adopters such as Accenture, Thomson Reuters, Peppermint, Templafy, and Petrobras are already leveraging the Retrieval API for a variety of enterprise AI use cases.

Key Features

• Secure and trustworthy: Results are permission-trimmed at query time, follow Microsoft Entra ID access controls, and carry sensitivity label metadata.
• Developer control with enterprise flexibility: Developers can use custom filters using metadata fields and connector labels for fine-grained control over retrieved content.

Use Cases

1. Thomson Reuters

• Thomson Reuters is integrating its AI assistant, CoCounsel, with Microsoft 365 via the Retrieval API to enhance legal workflows. CoCounsel allows legal professionals to query contracts stored in SharePoint or OneDrive for key clauses like indemnity or confidentiality.

2. Policy and Process Agents

• Build agents that retrieve relevant snippets from SharePoint and connector-indexed documents to answer questions like "What's our latest expense policy?" and provide up-to-date internal content.

3. RAG-based Applications

• Use the Retrieval API in retrieval-augmented generation (RAG) pipelines to ground model responses in tenant-specific content while ensuring compliance with organizational security policies.

Getting Started

• The Retrieval API is accessible via Microsoft Graph for public preview.
• Start with a POST request containing a natural language query and the targeted data source.
• Utilize filtering support, scalability (up to 25 documents per query), and batch support for querying multiple prompts or workflows in a single API call.

Conclusion

The Microsoft 365 Copilot Retrieval API empowers developers to unlock powerful AI capabilities by securely accessing and leveraging enterprise content. We are excited to see the innovative solutions that developers create with the Retrieval API and welcome feedback to enhance the API further.