Every new AI agent your team deploys starts from scratch: no memory of how the business works, where the data lives, or what rules apply. And as agentic coding tools spawn applications so fast that no one can control them, each one risks becoming another silo entirely outside your data layer. Microsoft is addressing both problems directly in Build 2026.
According to VentureBeat’s VB Pulse’s Q1 2026 RAG Infrastructure Market Tracker, hybrid recovery intent among organizations over 100 employees tripled from 10.3% in January to 33.3% in March, a sign that enterprises have moved beyond expanding RAG coverage and are now focusing on the architecture underpinning it. The shared business context is the part that recovery doesn’t solve.
On the context side, Microsoft is expanding its existing business data context layer Fabric IQ into a broader integrated system called Microsoft IQ, adding three additional context sources covering how the organization works, what it knows, and real-time global signals from the web, so any agent can use all four as a single base. On the application side, Refin, a new open-source SDK and CLI, deploys agent-built applications directly to the Fabric as a controlled production backend, routing application data into a single platform rather than spinning up new silos.
Microsoft Fabric CTO Amir Netz resorted to a movie analogy to explain where the data platform fits in. green screen of cascading code "math question" It was not the atmosphere, it was the layer that created the world in which Agent Smith operated.
"Our job in the data world is to create reality for agents based on data," Netz told VentureBeat.
Microsoft IQ unifies four reference sources into a single agent foundation
Microsoft IQ brings together four context sources that until now existed separately, designed in such a way that a developer can connect a new agent from all four in a single integration step.
Working intelligence. It shows how the organization works day-to-day, using emails, documents, meetings and schedules to give agents an understanding of people, teams and workflows.
Foundry IQ. Manages institutional knowledge, collecting and indexing knowledge bases so that agents can understand what it means to work within the organization, what rules apply, and what procedures to follow.
Fabric intelligence. Fabric models the live operational state of the business through defining data, entities, relationships and business rules based on real-time signals from real-time intelligence. Ontologies, the layer that captures that operational context, are expected to reach GA in the coming months.
Web IQ. Connects real-time global context from the web, giving agents a current picture of the organization’s internal data as well as the world outside the organization.
"Agents are going to become highly informed virtual employees," Netz said. "The world is going in this direction."
Refin routes agent-built applications to the same data foundation
Creating a shared reference solves half the problem. The second is when agents start building applications. Every new app needs a backend, and without a controlled deployment path every app creates a new data silo entirely outside the context layer.
Refin provides an enterprise-grade back end and deploys agent-built applications directly to Fabric, so application data lands in Microsoft OneLake by default and feeds back into the Microsoft IQ context layer instead of being stored outside of it.
Microsoft pits Refin against SupaBase and Neon, making Postgres-compatible backend agentive coding tools the default. The differentiator is governance: Refin routes the entire application fleet through Fabric’s unified data and compliance layer rather than creating separate silos.
Netz described the relationship as bidirectional. The agent that builds the Refine application takes inspiration from the organization’s ontology. The data that the application generates enriches that ontology for the next agent.
Every major data platform is chasing the same answer, but execution is unproven
Microsoft isn’t the only platform to create a shared context layer for agents. Snowflake announced its own contextual capabilities this week with semantic capabilities. Pinecone has its Nexus platform that extends vector databases to become a knowledge engine and Redis has developed its Iris context and memory platform.
Microsoft’s approach further reinforces the trend that RAG and model availability is no longer an issue.
"Fabric IQ and Refine are important because the enterprise AI challenge is no longer just about model availability," Robert Kramer, managing partner of KramerERP, told VentureBeat. "The real question is whether Microsoft simplifies execution and strengthens trust or adds another layer to an already complex environment."
<a href