geniosAI Raises $19.1 Million in Funding, Aiming to Create Traceable Working Papers for Financial Analysis

At earnings calls, when analysts are pressed for details on a particular figure, they need to provide the calculation process, data sources, and reasons for any adjustments. A well-written explanation alone is usually not enough. For financial AI to be adopted within institutions, it must first adapt to this way of working.

On November 9, 2026, geniosAI announced the end of its stealth mode and disclosed that it had raised a total of $19.1 million in seed and Series A funding from investors including FactSet, Zoom, and General Catalyst, with FactSet leading the Series A round. Headquartered in San Jose, California, and led by co-founder and CEO Prabhdeep (PD) Singh, the company serves banks, investment teams, and CFO offices.Company AnnouncementThe financing and business positioning were outlined.

Assumptions and audit opinions should also be retained.

geniosAI aims to attach sources, calculations, assumptions, and review decisions to analysis results so that teams can build upon them the next time they undertake similar work. Its product combines a knowledge framework for the financial sector, intelligent agents, and traceable task orchestration. FactSet, as a strategic partner, provides financial data support for this initiative.

This approach addresses a common handover challenge: the spreadsheet remains, but the person who created it has changed. Subsequent analysts can see the final figures, but they may not know why a particular expense was excluded or why a certain comparable company was included. When the rationale behind these decisions disappears along with the staff member, the next round of updates often requires explaining everything from scratch.

In the official website demo, a number can be expanded downward.

existProduct Examples on the Official WebsiteOn this page, readers can continue to view revenue, costs, one-time adjustments, and year-over-year comparisons based on adjusted EBITDA. The page also demonstrates that when the system encounters companies with the same name or missing financial statements, it prompts users to provide clarification rather than automatically filling in the blanks. These are official demonstrations of how the system works; they do not represent publicly audited client financial data and should not be construed as proof of accuracy for all real-world tasks.

Consider a quarterly update process: First, determine the subject of the research and the time period, then gather financial statements and comparable company data; systematically organize the changes and link the calculations to their sources; analysts review unusual items and confirm one-time adjustments; finally, use the approved figures in memos and presentation materials. The most important thing to document here is the reason for each revision, not just the version number of the final document.

If year-over-year data has been restated, the system should be able to indicate whether the old or new methodology was used. If an adjustment item changes, the system should prompt the user to review the relevant conclusions. These requirements are suitable as acceptance criteria for the trial period; however, the purchaser must still verify them using its own tasks and cannot rely solely on the official website demonstration to determine whether they have been met.

Traceability does not mean the conclusion is necessarily correct.

Even if every step of a calculation is properly sourced, the comparison may still be inappropriate. Differences in fiscal years between companies, variations in revenue recognition methods, or the grouping of one-time expenses of different natures can all lead to misleading results. Source links address the question of “where the information comes from,” but analysis and judgment must also determine whether “such a comparison is reasonable.”

Therefore, when evaluating such platforms, financial teams should prepare samples that are intentionally ambiguous: companies with the same name, different currencies, restated financial statements, missing data, and management-adjusted metrics. Observing when the system flags issues, how it explains conflicts, and whether user corrections affect related materials provides more valuable insights than simply reviewing a single report that was processed without a hitch.

Institutional buyers will ask how the data is isolated

geniosAI atSafety InstructionsThe document outlines commitments such as customer isolation, fine-grained permissions, customizable retention periods, audit logs, and the practice of not using customer data to train models. These are the vendor’s descriptions of its products and policies; specific deployment details, contractual obligations, and supporting documentation must still be confirmed during the procurement process.

Financial institutions should also ask: Which records will be retained after the data is deleted? What information will external model services have access to? Will auditors be able to view the complete source data? And does the data provider’s license cover the intended use? Before integrating non-public financial data into the system, it is more important to spend time clarifying these boundaries than to focus on the speed of data generation.

What makes this funding round noteworthy is that geniosAI has grounded its product value in the team’s sustainable working papers. The ability to retain analytical evidence and support challenges and revisions will determine whether it truly alleviates the burden on institutions. If auditors still have to reconstruct calculations item by item, then no matter how fast the automated generation is, it may simply be pushing the work down the line to the next stage.