Finpilot has raised $10 million in a Series A round to expand software that helps institutional investment teams handle document-heavy research and reporting. The Seattle company says its agents prepare information for people making investment decisions, rather than replacing their judgment.

The big change

  • The November 4 funding backs a product aimed at institutions that receive investment information across reports and disconnected systems.
  • Finpilot describes configured workflows that prepare evidence and stop for human review at specified decision points.
  • For firms assessing AI-assisted extraction, the value of a draft depends on whether staff can check it and manage errors.

Putting records to work

Finpilot’s announcement says Lobby Capital led the financing, with Madrona participating. The company plans to expand its product and team. It names institutions including Seattle Children’s and the Wyoming Retirement System as customers; it does not disclose the scope of each institution’s use.

The intended users include teams that allocate money across outside managers. Those managers send reports and spreadsheets whose contents have to be incorporated into investment records. In an interview with GeekWire, CEO Lakshay Chauhan said Finpilot brings data from portfolio systems and document storage together, matching funds, holdings and clients to identifiers.

Where a person enters the workflow

Finpilot’s product page illustrates a quarterly manager letter arriving in a data room. Its described workflow prepares an update against approved sources and existing context; a material change is held for the investment team. This is a company illustration of the intended process, not an independently observed customer run.

That distinction matters because preparing a summary and deciding what the change means for a portfolio are separate tasks. The page describes configurable sources, permissions and approval points, with a person able to approve, return an item with guidance or mark it immaterial. The example does not establish that every customer has the same rules.

Chauhan told GeekWire that a user could monitor updates from 35 private-equity managers and direct an analyst toward items needing attention. He also said the service links figures to their source documents and asks for approval when needed. Those are descriptions of how the company intends the system to work, not independently measured accuracy or time saved.

Some customers use the software alongside existing systems, while others have replaced older research or relationship-management tools, Chauhan told the publication. He reported roughly 28 institutional customers; the interview does not establish how extensively each uses the agents.

Checking the output

The wider oversight issue goes beyond producing a readable briefing. In its 2026 oversight report, FINRA describes summarization and extraction as common generative-AI uses among member firms and discusses validation, human review and error checks. That is general broker-dealer context, not a finding about Finpilot or a claim that the report governs every institution using it.

Source links and approval points support review, but do not establish that extraction is complete or correct. Independent accuracy and customer-level savings remain unverified.

Finpilot invites prospective customers to request a demonstration. Public material does not establish a general self-service launch, pricing or a universal approval policy. BIG CHANGE did not test the service.