Daloopa, a provider of structured financial data, has launched a new connector based on the Model Context Protocol (MCP) to feed its source-linked data into Microsoft 365 Copilot, the companies announced on June 25, 2026. The integration, unveiled in New York, aims to give finance professionals direct access to accurate, auditable financial figures within their everyday Copilot workflows without leaving Word, Excel, or Teams.
For the uninitiated, MCP is an open standard originally introduced by Anthropic that lets AI assistants securely tap into external tools and data sources. Microsoft adopted the protocol for its Copilot ecosystem earlier this year, and Daloopa’s move is among the first to bring finance-grade datasets—think company fundamentals, historical filings, and consensus estimates—into the conversational AI loop with full provenance tracking.
What the connector actually does
Behind the marketing speak, Daloopa’s MCP connector acts as a bridge between its cloud-based financial database and the Microsoft 365 apps where knowledge workers already spend their day. A financial analyst drafting an Excel report can type a natural language request like, “Pull Apple’s last five quarters of revenue and gross margin, and cite the source,” and Copilot fetches the numbers directly from Daloopa’s repository, complete with a link back to the original SEC filing or press release.
Every data point Daloopa serves is “source-linked,” meaning it’s tied to a specific line item in a public document. That matters in finance, where a misplaced decimal or a mismatched quarter can torpedo a model. By baking that reference chain into the MCP connector, Daloopa ensures Copilot’s outputs aren’t just plausible—they’re verifiable. Users can click through to the exact paragraph of an earnings transcript or the correct table in a 10-K, turning the AI from a black box into an auditable research assistant.
Why this matters for the Copilot ecosystem
Microsoft has been aggressively courting enterprise developers to build on top of Copilot, and MCP is central to that strategy. Think of it as a universal plug: any service that speaks MCP can become a Copilot extension, chipping away at the walled gardens that have long kept business data siloed. Daloopa’s connector demonstrates the model in a high-stakes domain. Finance teams are conservative about tooling—Excel is entrenched, and any AI that can’t show its work is a nonstarter.
The timing is also telling. Microsoft recently rolled out Copilot for Finance, a specialized role-based agent that automates reconciliation, variance analysis, and other grunt work. Connecting a robust financial data source like Daloopa’s turbocharges that agent. Instead of relying only on internal ERP numbers or manually pasted figures, Copilot can now cross-reference live market data, bringing external context inside the firm’s own spreadsheets.
The Model Context Protocol: a quick primer
For those who haven’t tracked the MCP story, the protocol solves a fundamental AI interoperability problem. Large language models are notoriously ignorant of what happened after their training cutoff and lack real-time access to proprietary systems. MCP gives them a standardized way to request information from a server—be it a database, an API, or a file system—without custom code for every integration.
Anthropic released MCP as an open standard in late 2024, and adoption has been swift. Microsoft incorporated it into Copilot Studio, allowing partners to build connectors once and deploy them across the entire Copilot surface: Outlook, Word, Excel, Teams, and the web. Daloopa’s connector is one of the first purpose-built for institutional finance, and it signals that MCP is mature enough for data where errors carry regulatory weight.
Daloopa’s dataset and what makes it special
Daloopa isn’t a household name outside niche finance circles, but within them it has a reputation for tackling one of the industry’s oldest headaches: data wrangling. The company automates the extraction and structuring of financial data from earnings releases, investor presentations, and regulatory filings, normalizing it into a consistent format. Crucially, it maintains every source link, so a user can always trace a figure back to its origin.
Before this connector, pulling that data into Copilot required clumsy workarounds—exporting CSVs, building custom APIs, or trusting the AI’s internal “knowledge” which is often stale and unattributed. Now, the entire Daloopa library becomes a native Copilot resource. The connector respects the existing permission model, so only users with a valid Daloopa subscription can access the data, preserving commercial boundaries.
Real-world workflows that become possible
Imagine a portfolio manager preparing for a morning meeting. She opens a Word document and asks Copilot to “Generate a summary of last night’s earnings from the five largest tech companies, include revenue surprises and forward guidance.” Within seconds, she gets a table populated with Daloopa-supplied figures, each hyperlinked to the source text. She can copy that into an email, and the links survive, so recipients can verify themselves.
In Excel, a quant could build a template that auto-refreshes with the latest consensus estimates every morning. “Update my financial model with Daloopa’s current estimates for Netflix, including subscriber growth and ARPU,” Copilot executes, pulling data through the MCP connector and populating the spreadsheet. Because the connector understands the schema of Daloopa’s data—not just raw text—it knows which columns map to what, reducing the risk of misalignment.
Even in Teams meetings, a CFO could summon live data without sharing a screen: “Copilot, what was Microsoft’s cloud revenue last quarter, and how did it change year-over-year?” The assistant answers orally, then drops the cited numbers into the meeting chat. It’s the kind of ambient intelligence that felt like science fiction just two years ago.
Industry reactions and early feedback
While Daloopa and Microsoft are still gathering telemetry from the early adopter program, the announcement has drawn positive murmurs from the financial technology community. Analysts who have seen demos describe the connector as “a missing puzzle piece” for Copilot’s finance ambitions. The ability to cross-check AI-generated claims against original filings addresses the trust gap that has held back generative AI in regulated industries.
That said, some skeptics note that no connector is a silver bullet. Data latency remains a concern: Daloopa’s machine-learning pipeline typically turns around earnings filings within hours, but for high-frequency traders, even minutes matter. The company says it’s working on near-real-time updates, though that’s a future deliverable. For the vast majority of corporate finance users, analysts, and portfolio managers, however, the current speed is more than adequate.
Competitive landscape and what’s next
Daloopa isn’t alone in the race to plug financial data into AI copilots. Bloomberg and Refinitiv have both teased natural language interfaces, and several startups are building MCP servers for alternative data. What sets Daloopa apart is its focus on source-linking—a feature that is expensive to implement and maintain but increasingly seen as table stakes for any data provider that wants a seat at the enterprise AI table.
Microsoft, for its part, is likely to highlight this connector at upcoming events as proof that Copilot is ready for serious business functions. Expect more vertical-specific MCP connectors in healthcare, legal, and supply chain soon. The pattern is clear: instead of forcing customers to bring their data to the AI, Microsoft wants the AI to reach into the tools where data already lives.
For Daloopa, the connector opens a new distribution channel. Every organization that standardizes on Microsoft 365 becomes a potential customer. The company has started offering bundled trials that include the connector and a starter data package, with pricing scaling based on the breadth of coverage. Early adopters report that the integration slashed the time spent on manual data entry by up to 70% in pilot studies, though those figures are self-reported.
How to get started
Existing Daloopa subscribers can activate the MCP connector through the Copilot Studio admin panel. IT admins must approve the connector and map it to the appropriate security group, a step that usually takes less than ten minutes. Once activated, users simply select Daloopa as a data source within any Copilot-enabled app and authenticate once with their Daloopa credentials.
Microsoft has published a detailed setup guide on its documentation site, and Daloopa’s support team is offering white-glove onboarding for the first 100 enterprise customers. For firms that haven’t yet adopted Copilot, Microsoft’s field sellers are packaging the connector as part of a broader “AI for Finance” bundle that includes Copilot for Microsoft 365 and the Daloopa data subscription.
Broader implications for the Windows ecosystem
Although the connector lives at the intersection of cloud services and Office apps, it carries weight for Windows users. Microsoft 365 Copilot is deeply integrated into Windows 11, with a system-wide sidebar and the ability to interact with files and apps. As more MCP connectors come online, Windows itself becomes a more powerful AI orchestration layer. A financial analyst’s Windows desktop could soon be the central command post for workflows that pull data from dozens of specialized sources, all mediated by natural language.
This trend also dovetails with the upcoming Windows 12 release, which is expected to bake Copilot even deeper into the OS. While Microsoft hasn’t officially confirmed MCP as a local protocol on Windows, developers have spotted references in preview builds that suggest a future where desktop apps can also expose MCP servers. If that materializes, a connector like Daloopa’s wouldn’t just live in the cloud—it could run directly on a finance PC, querying local data alongside cloud sources.
Looking ahead
The Daloopa connector is a concrete example of how MCP is turning AI copilots from generic assistants into domain-specific power tools. For finance teams, the immediate benefit is speed and accuracy; for Microsoft, it’s a validation of the open-connector model. The real test will come when auditors, regulators, and large asset managers begin to rely on these integrations for mission-critical decisions. If the source-linking holds up under that scrutiny, expect a stampede of financial data vendors to follow Daloopa’s lead.
In the meantime, the June 25 launch is a signal that the Copilot ecosystem is maturing beyond simple productivity hacks. The next wave of connectors won’t just look up calendar appointments or summarize documents. They’ll bring hard data—the kind that moves markets—into every conversation, with a level of trust that was previously unattainable in an AI context. For Windows users in finance, the future just became a little more automated and a lot more verifiable.