On June 25, 2026, financial data platform Daloopa announced a Model Context Protocol (MCP) connector for Microsoft 365 Copilot, giving investment professionals direct access to its structured, source-linked data on more than 5,500 global companies from within Word, Excel, PowerPoint, and other Microsoft 365 apps. The integration represents one of the first financial-sector implementations of MCP—an open standard pioneered by Anthropic and rapidly adopted by Microsoft for extending Copilot’s capabilities—and promises to slash the time analysts spend on manual data retrieval and verification while reducing the risk of AI-generated hallucinations.
Daloopa is already a staple on Wall Street, known for building machine-readable data sets directly from financial filings, earnings call transcripts, and regulatory documents. Each data point is linked back to its original source, a feature that becomes uniquely powerful inside a generative AI environment where traceability is critical. With the MCP connector, Copilot users can query Daloopa’s database using natural language and receive answers complete with citations, turning an AI assistant into a research-grade analytics tool.
“Our mission has always been to eliminate the grunt work in financial analysis—the copying, pasting, and manual checks that eat up 30% of an analyst’s day,” said Dhruv Chopra, CEO of Daloopa, in an interview. “By making our data accessible through Copilot via MCP, we’re putting a trusted, audit-ready layer directly into the workflow where decisions are made. Every number can be traced back to the 10-Q or earnings transcript it came from.”
The move aligns with Microsoft’s broader push to turn Microsoft 365 Copilot into an extensible platform. Since announcing Copilot extensibility at Build 2025, Microsoft has onboarded dozens of data providers and line-of-business applications via MCP connectors. The protocol, originally developed by Anthropic for its Claude model, defines a standardized way for AI models to communicate with external tools and services. Microsoft adopted it in early 2026 as the backbone for Copilot’s new agent framework, replacing the earlier Graph-based connectors for third-party data.
What the Daloopa MCP Connector Unlocks
When an investment professional activates the Daloopa connector in their Copilot settings, a new data source becomes available across all M365 applications. In Excel, a user could type “Pull the last 8 quarters of revenue for Apple, Microsoft, and Nvidia, and compare their growth rates” and Copilot would retrieve the exact, source-linked figures from Daloopa’s database, creating a formatted table with footnotes pointing to specific filing pages. In Word, an analyst drafting a research note could ask Copilot to insert a paragraph summarizing a company’s latest capital allocation trends, with the AI synthesizing multiple Daloopa data points and automatically generating citations.
Crucially, Copilot’s existing ability to interact with unverified web data is replaced, for Daloopa-sourced queries, by a deterministic pull from the platform’s structured repository. This dramatically reduces the risk of hallucination—a persistent problem in financial AI. Early tests showed Copilot’s error rate on financial metrics dropping below 1% when anchored to Daloopa’s source-linked data, compared to a baseline of 8-12% when relying on web-scraped information.
“What’s different here is the marriage of a curated, proprietary dataset with the general reasoning capabilities of a large language model,” said Ravi Mehra, a financial technology consultant who beta-tested the integration. “Copilot isn’t just summarizing text; it’s doing actual quantitative analysis on top of verified numbers. That’s a huge leap from the first generation of AI assistants.”
How Source-Linking Defends Against Hallucination
Daloopa’s core differentiator has always been its “source-linked” methodology. When a company releases earnings, Daloopa’s AI models and human validators break the document into standardized line items—revenue, gross margin, EPS, etc.—and map each to the exact sentence or table cell from which it came. That link is preserved as immutable metadata. The MCP connector exposes these links to Copilot, which can then cite them inline. Users can click any number to see a popup with the original source snippet, a feature that Daloopa’s enterprise clients have relied on for years in their own internal tools, now brought natively into a general-purpose productivity suite.
This traceability addresses a growing regulatory concern. The SEC and FINRA have signaled that AI-generated investment recommendations must be backed by transparent and auditable data sources. In a 2025 risk alert, FINRA warned that “firms using generative AI for client communications or research must ensure that outputs are attributable to specific, reliable data.” Daloopa’s integration effectively gives compliance teams a way to verify every Copilot-generated figure at the atomic level, potentially shaving weeks off internal review cycles for research reports and client presentations.
Moreover, because the data lives inside Daloopa’s secure cloud—not in Copilot’s general training corpus—financial institutions can govern access with the same granular controls they apply to any other sensitive dataset. Microsoft 365’s built-in data governance tools, including sensitivity labels and data loss prevention policies, apply to Copilot queries that tap the Daloopa connector, allowing firms to restrict which analysts can pull which companies’ data, enforce geographic boundaries, and log every request for audit.
The Model Context Protocol’s Rapid Expansion within Microsoft 365
The Daloopa launch is one of a flurry of MCP-based integrations hitting the Copilot ecosystem. In May 2026, Microsoft CEO Satya Nadella announced that more than 200 MCP connectors were available in the Copilot catalog, spanning CRM systems, project-management tools, and specialized databases. The protocol’s simplicity—a JSON-based schema that describes tool capabilities and authentication methods—has allowed niche data providers to build Copilot integrations in weeks rather than months.
For Daloopa, the decision to build on MCP was strategic. “We considered building a custom Copilot plugin using the older Graph connector approach, but MCP gives us a future-proof standard that works not just with Microsoft 365 but potentially with other AI platforms that adopt the protocol,” said CTO Angela Chen in a blog post. That multi-platform potential is significant; Anthropic’s Claude and other models could eventually tap the same Daloopa endpoint, making the engineering investment go further.
Microsoft’s own embrace of MCP marks a departure from its historical preference for proprietary extensibility models. Analysts see it as part of a broader effort to position Copilot as the AI-agnostic orchestrator of enterprise workflows. “By betting on MCP, Microsoft is acknowledging that no single AI model will own enterprise data access,” said Forrester analyst Peter Wallace. “It’s an open door for best-of-breed data providers like Daloopa to plug in and compete on the quality of their data, not their AI partner.”
Real-World Workflows for Investment Teams
Early adopters described a variety of use cases. A global macro hedge fund is using the connector to automatically update its weekly portfolio risk reports. Previously, two junior analysts spent every Friday morning collecting and verifying financial metrics from a watchlist of 200 companies. Now, a script in Excel with Copilot queries Daloopa, populates the sheet, and flags any values that deviate more than 5% from the prior week—a process that takes nine minutes.
A sell-side equity research team set up a Word template where Copilot generates a first-draft earnings summary for tracked companies within minutes of a filing hitting Daloopa’s system. “We used to wait 24 hours for our internal data team to process a 10-K,” said the team’s head of research. “Now I can have a draft with verified numbers and source links before the company’s conference call starts. It doesn’t replace judgment, but it eliminates the mechanical part.”
PowerPoint users are leveraging the connector to create client-ready charts with an audit trail. A request like “Create a slide showing Tesla’s free cash flow trend over the last three years, citing the sources” produces a graph with footnotes that link directly to the relevant SEC filing pages—a feature that drastically reduces the back-and-forth between analysts and compliance.
Governance, Security, and the Compliance Landscape
For heavily regulated financial institutions, data governance isn’t optional. Daloopa and Microsoft have jointly published architecture documentation showing how data flows through the connector. Queries are processed via Azure-hosted MCP endpoints; Daloopa’s API returns only the specific data requested, and Copilot does not cache or train on that data. Role-based access controls can be configured through Microsoft Entra ID, and all interactions are logged in Microsoft Purview for compliance monitoring.
Daloopa also announced that its connector supports data residency requirements, allowing firms to designate that financial data be processed in specific Azure regions. This was a key unlock for several European and Asian investment banks that had previously been unable to use Daloopa with cloud-based AI tools due to local data sovereignty laws.
“Data governance was the elephant in the room for financial AI,” said Mehra. “Most funds were using public-ChatGPT workarounds with obvious risks. This integration puts the data on a scaffold that boards and regulators can actually inspect. That’s the breakthrough.”
Still, some compliance officers remain cautious. A senior attorney at a large asset manager noted that while source-linking helps, it doesn’t fully remove the burden of “reasonable basis” under FINRA rules. “If a model misinterprets a source—say it picks the wrong segment revenue line—the citation still points to the source, but the number could be wrong. Firms need human review processes around any AI output, linked or not.”
Pricing and Availability
Daloopa’s MCP connector is available starting June 25, 2026, to all Daloopa enterprise customers as part of their existing subscription, with no additional licensing fee for the connector itself. Users will need a Microsoft 365 Copilot license and must enable the connector via the Microsoft 365 admin center. Daloopa said it will offer a free trial tier for the connector through September 2026, providing access to a limited universe of 500 companies to let teams test the integration.
For non-Daloopa customers, the company has introduced a new “Copilot Essentials” tier at $15,000 per seat per year, which includes access to Daloopa’s core database—5,500+ companies, historical data going back 20 years—and the MCP connector. That price point puts it in reach of smaller hedge funds and boutique advisory shops, a departure from Daloopa’s historically enterprise-only model.
Industry Reactions and Competitive Dynamics
The announcement drew immediate attention from the financial-data ecosystem. Rival platforms like AlphaSense and Sentieo have their own Copilot integrations, but none yet offer MCP-based source-linking with the same depth of company coverage. Bloomberg, which continues to rely on its proprietary terminal ecosystem, has not publicly committed to MCP, though executives have hinted at future AI integration plans. “This puts pressure on every data provider to either adopt MCP or risk becoming a walled garden that AI assistants can’t access,” said Wallace.
Some analysts warned that Copilot’s adoption on Wall Street is still nascent. Many firms are only now rolling out the AI assistant to non-investment staff, and concerns about data leakage have slowed deployment in front-office roles. However, integrations like Daloopa’s could accelerate adoption by demonstrating a clear, measurable ROI. A recent survey by Greenwich Associates found that 62% of buy-side firms expect to deploy AI tools linked to proprietary or curated datasets by mid-2027.
A Glimpse Ahead
Daloopa and Microsoft plan to deepen the integration in future updates. A roadmap item for Q4 2026 includes “proactive narrative generation,” where Copilot, using Daloopa data, automatically drafts a short paragraph summarizing material changes in a company’s filings when they occur, pushed directly to a user’s Outlook inbox. The companies are also exploring a Teams integration that would allow an analyst to @mention Copilot during a meeting and ask it to pull real-time financial data from Daloopa for a company being discussed.
On the Daloopa side, the company is expanding its dataset beyond pure financial metrics to include ESG scores, corporate governance indicators, and operational KPIs—all source-linked to disclosures and third-party ratings. That data will also be exposed via the MCP connector, turning Copilot into a multi-dimensional research assistant.
Microsoft, for its part, continues to aggressively build out MCP support. A source familiar with the matter said that Microsoft is developing a “data governance layer” for MCP that will allow administrators to set cross-connector policies, such as blocking all connectors from accessing non-public financial data unless a user is explicitly authorized. This would address the most common complaint from IT security chiefs at financial institutions.
As AI becomes embedded in the tools of finance, the quality and traceability of underlying data will separate serious institutional platforms from consumer-grade experiments. Daloopa’s MCP connector is an early signal that the era of hallucinated financial analysis may be ending—replaced by a future where every AI-generated number comes with a receipt.