Dun & Bradstreet Finance Analytics Connects Credit Data to Microsoft Copilot via MCP

What was announced​

Dun & Bradstreet (D&B) announced the D&B.AI capabilities for its Finance Analytics platform on September 24, 2026. The company says the goal is to help finance teams make faster credit decisions, surface portfolio risks earlier, and optimize growth opportunities.

There are two ways to use them:

• Inside D&B Finance Analytics. Users can access D&B.AI capabilities for company research, risk analysis and guidance, identity verification, and portfolio management and monitoring via a conversational AI agent embedded within the D&B Finance Analytics platform.
• Through MCP. Users can also extend D&B business context and credit intelligence into AI platforms and enterprise applications through MCP connectivity to create custom agentic workflows.

Here is where it works today and what is still coming. D&B says the capabilities are available today in-platform and in Claude, Codex, ChatGPT, Microsoft Copilot, and Databricks via MCP server. MCP integration is also planned for Gemini Enterprise. Gemini support is planned, not shipping, so don't count on it yet.

Section summary: D&B has added a chat agent to its own finance product and made the same credit data available to outside AI assistants, including Microsoft Copilot, through MCP.

The data behind it: the Commercial Graph and D-U-N-S​

Underneath all of this is what D&B calls the Commercial Graph. Anchored by the global standard D‑U‑N‑S Number identifier, the D&B Commercial Graph provides a context layer that allows AI agents to understand business identity, relationships and risk across the global economy. D&B's own background material says it originated the D-U-N-S® Number in 1963, now the global standard for identifying commercial entities.

This matters because of how language models fail. A general model asked whether "Acme Holdings" is a good credit risk will usually give a confident answer, and it may be describing a different Acme. A fixed entity identifier gives the agent a firm reference point. The Distilled Funding newsletter put the practical meaning well: organizations with the appropriate D&B access can now connect AI agents directly to governed commercial-credit intelligence instead of asking a language model to guess from whatever it can find on the public internet. The same publication also points out that this does not mean every person who opens ChatGPT can suddenly pull your PAYDEX score. You need a D&B relationship to get access.

D&B's release also says Commercial Graph outputs are "consistent, traceable, and auditable." That is the company's description. No independent party has verified it.

What D&B says about results​

D&B's headline number is a credit analysis and research accelerated by up to 30–40% compared with industry benchmarks for organizations using the features in Finance Analytics. The company's release makes two more claims:

• Credit losses cut by up to 20–25%, which D&B credits to earlier warning of emerging risk signals.
• Growth opportunities up by up to 10–15%, which D&B credits to better portfolio insight and prioritization.

All three figures should be read with caution. The release does not say which industry benchmarks were used. It also gives no sample size, time period, customer mix or method. These are best-case "up to" numbers from the vendor, not guaranteed or independently checked results. The same goes for the claim by Scott Spencer, D&B's general manager for finance and credit, that teams get trusted answers "in seconds instead of hours." It is a quote from a launch, not a service-level commitment. Spencer described the problem this way: "Credit teams are under pressure to make faster decisions without sacrificing confidence, consistency or governance."

Section summary: The performance numbers are vendor claims with no disclosed method. Use them to shape your own pilot, not as proof.

Why this matters for Microsoft shops​

This is D&B's third announcement in four months about putting its data inside other companies' AI tools, and the second tied to Microsoft in two weeks.

• June 2026, OpenAI. D&B said customers could securely connect D&B-hosted MCP servers directly to ChatGPT and Codex. It also said Finance Analytics tools were reachable via MCP for automated business credit decisions powered by real-time insight into hundreds of millions of businesses and a rules-based engine.
• September 16, 2026, Copilot Studio and Dynamics 365. D&B announced new Model Context Protocol (MCP) integrations with Microsoft Copilot Studio and Microsoft Dynamics 365 Sales agents. The integration lets users enrich Microsoft's prebuilt Dynamics 365 Sales agents, such as the Sales Qualification Agent and Sales Opportunity Agent, with D&B's continually verified business context. In Copilot Studio, D&B says custom agents can use the graph for workflows such as credit and risk evaluation, third-party onboarding, compliance, supply chain management, and account-based marketing.
• September 24, 2026, Finance Analytics AI features. This release names Microsoft Copilot as a place where the finance features are available through MCP.

Keep the September 16 and September 24 announcements separate. The Copilot Studio and Dynamics 365 integrations were announced first and on their own. The Finance Analytics release did not introduce them.

D&B's AI Connectors page also lists a third Microsoft offering: a Microsoft Copilot Graph Connector. D&B describes it as a no-cost, curated sample of basic business data covering tens of thousands of public and private companies, and says a Copilot administrator must enable it. Don't assume this free sample connector covers the same data, scope or cost as the Finance Analytics MCP features. D&B presents them as different products.

The general pattern is clear, though. MCP has become the standard way for data vendors to reach every AI assistant at once. Rather than building five separate integrations, D&B runs one MCP server. The same page says it can be connected to MCP-compatible runtimes with a single API key, and customers license selected "AI-ready" D&B Data Blocks. MCP works like a universal power adapter: the AI platform is the socket, and D&B is one of many appliances that now fit it.

Section summary: For Microsoft customers there are now three D&B routes: Finance Analytics features through MCP in Copilot, Copilot Studio and Dynamics 365 Sales agent integrations, and a free sample Graph Connector. Each has its own scope.

A checklist for IT and security teams​

The announcement does not say which account types or product tiers are supported. It also leaves out regional availability, pricing, setup steps, and details on permissions, logging or data retention. The points below are general guidance for handling any third-party MCP data source. They are not documented D&B procedures.

1. Confirm licensing first. Ask D&B which Finance Analytics subscription and Data Blocks cover MCP access, and whether usage is billed per call.
2. Find out which Copilot surface is involved. "Microsoft Copilot" in the release, Copilot Studio, Dynamics 365 Sales agents and the Graph Connector may each need different admin roles and licenses on the Microsoft side.
3. Map the data flow. Work out what leaves your tenant. A credit agent's prompts may contain customer names, internal credit policy or exposure figures. Get D&B and Microsoft to explain in writing how each side processes and keeps that data.
4. Scope who can use it. Credit data is sensitive. Limit which agents and which users can call the D&B tools, and apply your usual connector governance and data loss prevention policies.
5. Check the audit claims yourself. D&B describes its outputs as traceable and auditable. Before any agent helps set credit limits, test whether your own logs can show which D&B data drove a given recommendation.
6. Keep a human in the loop. Wherever a decision has regulatory or financial weight, have an agent's suggestion reviewed rather than acted on automatically, at least until your pilot shows how reliable it is.

The bottom line​

This launch is aimed mainly at credit and finance departments, but it lands on the IT team's desk because it runs through Copilot and MCP. It is useful news that verified business data can now reach Microsoft Copilot, alongside ChatGPT, Claude, Codex and Databricks, and that Gemini Enterprise is next. The value depends on details D&B has not published yet: licensing, data handling, and whether the "up to 40%" speedup holds up in your own portfolio. Treat it as something worth piloting, not something to trust blindly. As with any AI agent, keep your data governance in place before you let it make decisions.