Microsoft used its latest Build announcements to outline a broader push to make enterprise data and applications more usable for AI agents. The company introduced four related context layers, called Work IQ, Fabric IQ, Foundry IQ, and Web IQ, alongside new database and application tooling designed to help developers move AI projects closer to production.

The company’s central argument is that the next challenge in AI is no longer model capability, but context. Microsoft said workers and developers are increasingly handing off complete tasks to agents and coordinating multiple agents at once. That shift, it said, requires a shared understanding of how a business operates, where information lives, and what rules apply. Without that foundation, Microsoft argued, each agent begins from scratch and cannot reliably coordinate with others at scale.

To address that gap, Microsoft is positioning Microsoft Fabric as a unified data and AI platform that can provide shared context across an organization. The company said this foundation is intended to help businesses move from isolated AI experiments to production systems where agents can build on common data, semantics, and governance.

One of the new developer tools is Rayfin, an open-source SDK and command-line interface aimed at helping users turn prompts into production backends. Microsoft said Rayfin is designed to let developers and coding agents define what they want to build and generate enterprise-grade backend components such as databases, authentication, and application logic. It then deploys into Microsoft Fabric, which Microsoft says gives applications security and scale from the start. The company also said Rayfin uses GitHub-based workflows and stores application data in OneLake, making it available across Fabric’s analytics, operational, and AI layers.

Microsoft also announced Azure HorizonDB, a managed PostgreSQL-compatible database that is now in public preview. The company described it as a cloud-scale option built for AI-powered applications, with zone resilience by default, elastic storage up to 128 TB, and compute scaling to 3,072 vCores. Microsoft said the database is designed to support demanding transactional workloads with sub-millisecond multi-zone commit latency.

Beyond core database performance, HorizonDB includes AI-oriented features such as vector search, integrated AI model management, and direct connectivity to Microsoft Foundry and Fabric. Microsoft said these additions are meant to simplify architecture for teams building modern applications that need both transactional and AI capabilities.

The company also highlighted updates to its broader PostgreSQL portfolio. Microsoft Defender for Cloud integration for Azure Database for PostgreSQL is in preview, offering continuous security and compliance checks. New discovery and assessment tools are also available to help organizations plan migrations from Oracle and PostgreSQL environments, with readiness, sizing, and cost guidance.

For large-scale AI and multi-agent systems, Microsoft pointed to Azure Cosmos DB, where the Linux Emulator is now generally available across Linux, macOS, and Windows. New preview features include semantic reranking for search and an agent memory toolkit that works with Azure Cosmos DB, Azure Durable Functions, and Microsoft Foundry models.

The new IQ framework is the most explicit sign yet of Microsoft’s effort to standardize context for AI agents. Work IQ is designed to capture how work happens, Fabric IQ models how the business operates, Foundry IQ helps agents discover and reuse knowledge, and Web IQ adds live context from the internet. Microsoft said the goal is to give agents a shared foundation so they can reason consistently and act within governed business processes rather than as isolated tools.