Building the Data Layer for Agentic Finance: Our Investment in Daloopa

Much of today’s market discussion around AI in finance centers on agents displacing tedious, manual analyst workflows. While the promise is evident, in finance, “close enough” is not good enough. A misread footnote, misinterpreted KPI, stale segment breakout, or missed restatement can distort an insight leading to a poor investment decision and compliance issues. For AI-enabled tooling to reach its full potential, it needs more than better prompts and bigger models. It needs a data foundation built for accuracy, traceability, and domain-specific financial context. 

That belief sits at the core of our investment thesis in Daloopa. Our conviction starts with Thomas Li, Daloopa’s Founder and CEO, and his co-founders, who brought together firsthand experience from leading hedge funds and technology companies to solve a workflow problem they knew intimately: the persistent difficulty of collecting, structuring, and trusting public company data at scale.

Daloopa has built what we believe is one of the most important infrastructure layers for AI-enabled finance: structured, source-linked, as-reported financial data delivered in near real-time to power analyst workflows, models, and AI agents with confidence. The company covers more than 5,500 public companies globally, delivers up to 10 times more data points per company than other providers, and links every datapoint back to its original source for auditability.

Daloopa addresses the “last mile” complexity of public company data through proprietary, finance-specific extraction and normalization models. The result is a data accuracy flywheel that has compounded over time, giving them the data extraction capabilities to deliver the quality, depth, and speed required by many of the world’s leading hedge funds, asset managers, investment banks, and AI lab partners. Daloopa has also embedded itself deeply into both existing and next-generation workflows, delivering data where analysts already work: through its Excel updater, web portal, API, MCP connector, and proprietary Scout agent. 

Customer feedback consistently reinforced the same themes: Daloopa improves speed, accuracy, and confidence, while allowing teams to spend more time on judgment, interpretation, and decision-making.

Beyond the data asset and distribution, our thesis is driven by the belief that AI-enabled financial analysis will dramatically increase the value of Daloopa’s offering and broaden its addressable market.

Financial AI faces two data problems, the first being the obvious one: sourcing the wrong numbers. This remains difficult for general-purpose LLMs and web-based retrieval systems, given the source material is multimodal, inconsistent, and constantly changing. The second problem may be even larger: missing or misinterpreting numbers. In financial analysis, an answer is often not wrong because a model pulled the wrong figure, but because it failed to collect a datapoint (e.g. via connecting management commentary to a KPI trend) or misinterpreted a correct number (e.g. given a segment reclassification).

These errors of omission, context, and comprehension are harder to catch and become even more consequential as analysts rely on AI agents to surface answers. A missed KPI can materially change the conclusion an agent produces.

This is where Daloopa’s depth becomes strategically important. The company does not simply collect a narrow set of standardized financials, rather, it strives to capture every reported metric for a given company, including KPIs, guidance, reconciliations, and qualitative context across thousands of companies. That breadth gives AI systems more complete raw material to reason over.

Just as importantly, Daloopa’s data model creates an ontology that bridges written and verbal disclosures, company-specific definitions, and cross-company industry comparisons across disparate data types. As AI moves deeper into financial services, agents require the right context to update a model, compare reported KPIs to expectations, identify revisions, explain variances, summarize management commentary, draft an earnings reaction, and assemble research-ready artifacts. Those outputs are only as good as the data foundation beneath them.

Daloopa is positioned to deliver that next layer because it combines three things that are difficult to replicate: a trusted data asset, a finance-specific ontology, and deep workflow embedding. Instead of asking an AI model to repeatedly search, retrieve, clean, reconcile, and reason from scratch, Daloopa enables analysis on top of canonical, source-linked financial data. We believe that we can improve output quality, reduce analyst toil, and lower compute and token costs over time by giving AI systems structured context up front.

At Brighton Park, we look for exceptional teams, mission-critical products, and companies with the potential to define large markets. Daloopa fits that profile. Thomas and the Daloopa team have built a platform that solves a painful problem today and becomes even more strategic as AI adoption accelerates, and we are thrilled to partner with Daloopa for this next chapter of growth.