An agentic application is an AI system that knows your business context, reasons autonomously, and takes action based on real-time data and specialized expertise. Agent Bricks, Genie, Apps, and Lakebase give enterprises the tools to build agentic applications at scale. But there's a critical gap: agents built solely on internal data can't truly think.

Consider a loan approval agent. It has access to your bank's loan book, customer history, and credit scores. But it lacks the context that humans instinctively use:

• Current market conditions and sector trends
• Real-time creditworthiness signals (updated ratings, outlook changes)
• Property valuations and equity shifts
• Competitor activity and economic headwinds

Without this real-time intelligence, agents become knowledge-limited—constrained by historical data, unable to reason about the world as it is now. They can execute workflows, but they can't make informed decisions.

The old solution? Manual research. Analysts pull data from multiple sources, lose context switching between tools, and create bottlenecks. Decisions slow down. Risk increases.

Agents need a way to access live, trusted intelligence while they reason through complex problems. That's where the MCP Marketplace comes in.

The Solution: Real-Time Intelligence for Agentic Applications

The Databricks Marketplace is an open marketplace for data, analytics and AI. It is the central hub for discovering, connecting to, and governing MCP servers and AI-ready data, enabling agents to access external data at scale with minimal friction.

Let's walk through three domain-specific examples that illustrate how agentic applications use real-time intelligence to solve concrete problems. These three providers represent different types of external intelligence your agents might need:

• You.com — Real-time web intelligence and market sentiment
• Moody's — Institutional credit research and entity intelligence
• Cotality — Real estate and mortgage expertise for lenders

These aren't just data feeds. They're executable intelligence sources that agents can query, evaluate, and act upon, all within a governed, auditable environment. Every connection is authenticated through Unity Catalog, creating a single source of truth for access control and compliance.

By connecting your agentic applications to these MCP servers, you're doing three critical things:

1. Bridging the knowledge gap. Agentic apps access live data that wasn't in your training set or data warehouse.
2. Eliminating manual research. Workflows that once required human intervention now run autonomously.
3. Staying governed. Everything happens inside the Lakehouse, with full lineage and compliance tracking.

How It Works: Three Providers in Action

To illustrate how agentic applications leverage real-time intelligence, here's how these three providers power autonomous decision-making across different use cases:

You.com: Real-Time Web Context and Market Sentiment

The You.com MCP acts as a real-time research arm that can be invoked within your agents to provide the "why" behind internal performance metrics.

• The gap: Your dashboard shows a 10% regional sales dip in EMEA business. Why?
• Agent’s action: Your agent calls You.com MCP to retrieve live market news, weather patterns, competitor pricing, and regulatory changes. Your agent synthesizes the context.
• What agents access: You.com allows agents to pull real-time news, regulatory shifts, and competitor pricing directly into a Databricks Genie conversation to provide instant, grounded research.

This demo illustrates how a supervisor agent uses Genie for internal data warehouse queries and the You.com MCP Server for external web research.

Moody’s: Institutional Credit Research at Scale

Moody’s MCP Server delivers trusted, GenAI-ready intelligence that can be woven into financial applications to accelerate decision-making. By invoking Moody’s curated data ecosystem, including global entity ownership, financial statements, and Moody's Ratings credit ratings and research alongside your bank’s private loan book, you move from manual research to decision-ready intelligence.

• The gap: Your loan officer approves applicants based on credit scores alone. They're missing the full picture.
• Agent’s Action: Moody's provides current Moody's Ratings credit ratings, historical trends, peer benchmarks, and sector outlook. Your agent now compares the applicant against investment-grade peers, identifies hidden risks, and surfaces the rationale to your credit officer in seconds.
• What agents access: Business users can perform high-level entity searches, access current credit ratings and outlook history dating back to 2000, and retrieve forward-looking sector trajectories to support strategic planning.

Click here for the detailed walkthrough of Moody's MCP Server on Databricks

Cotality: The "Property 360" for Mortgage Excellence

The Cotality MCP and AI-Ready datasets provide the missing link for lenders, allowing them to embed real estate universe intelligence directly into their agentic workflows.

• The gap: Portfolio "leakage" often happens in the dark. Without a clear, unified view across data sources into recapture performance, you don't know if you’re losing a borrower to a new home purchase or a winnable refinance until the payoff request hits your desk.
• Agent's action: Genie scans your portfolio to resolve property data records and identify exactly where your book is going. By merging internal data with external signals, it distinguishes between move-up buyers and "Refi Shops" poaching your equity, allowing you to strengthen your strategy where it's most winnable.
• What Agents access: Cotality’s CLIP MCP–powered property resolution, paired with AI‑ready datasets, provides a live window into who is capturing your book and why. Originators can see if a payoff was a new purchase or a refinance, track competitor poaching in real-time, and deploy targeted outreach to stop run-off before it starts.

How Agents Remember: Lakebase as State Engine

Real-time intelligence solves half the problem. But agentic applications need another capability: persistent memory and state.

Here's the challenge: a complex workflow—like a 3-day commercial loan review—involves multiple steps, human handoffs, and decisions. Without a persistent state, agents restart from scratch each time. Context is lost. Progress resets.

That's where Lakebase comes into play. Lakebase is a serverless, stateful database purpose-built for agentic applications. It stores agent state, decisions, and audit trails throughout multi-step workflows.

Why this matters for real-time intelligence: When your agent accesses Moody's data on day one and Cotality data on day two, Lakebase connects those insights and reasons across them. The agent doesn't re-request data; it builds on what it knows. Every decision is logged for compliance—automatically.

Putting It Together: The Commercial Loan Decision Support App

Let’s see how a production-grade agentic application works

1. Application arrives → Lakebase stores status and metadata
2. Agent awakens → Retrieves context from Lakebase
3. Agent queries intelligence:
   • Moody's on Marketplace: Current credit rating, peer benchmarks, risk signals
   • Cotality on Marketplace: Property valuation, lien status, market trends
   • You.com on Marketplace: Regional economic outlook, sector momentum
4. Agent synthesizes → Combines internal data (your loan book) with external intelligence
5. Agent surfaces decision → Proposes approval/denial with full reasoning via Genie
6. Credit officer reviews → One-click approval in your Databricks App
7. Lakebase logs decision → Complete audit trail (decision, data sources, timestamp, approver) for compliance

Governance throughout: All data accessed, all agent reasoning, all decisions—tracked in Unity Catalog with full lineage.

Get Started: Build Smarter Agents Today

The MCP Marketplace is live. Real-time intelligence is now accessible to your agents:

Three providers are ready:

• You.com for market intelligence
• Moody's for credit and entity research
• Cotality for real estate expertise

Three next steps:

1. Explore the Databricks Marketplace to see all available MCP servers
2. Build your first agent with Agent Bricks, grounded in real-time data
3. Deploy with Lakebase for memory, governance, and auditability

Agents powered by real-time intelligence don't just execute workflows—they reason, adapt, and make autonomous decisions that drive business results.