Finpilot

• Turning a quarter's manager letters and capital account statements into cited sections of the institution's quarterly report
• Drafting the next investment committee book in the office's template, with flagged managers and open questions surfaced for the CIO
• Matching an incoming capital call notice to the right commitment and routing it for approval before it is booked
• Reconciling NAVs and distributions against the fund administrator's file and reviewing only the exceptions
• Watching every manager between quarters and flagging strategy drift or a key-person change with the passage that shows it
• Preparing an advisor's household brief with holdings changes, prior conversations and the firm's approved investment view before a client meeting

Finpilot is aimed at a specific, underserved job: the allocator's back and middle office, where a small team spends each quarter reading PDFs from dozens of managers and administrators, rekeying numbers and assembling committee books. Its design choices match that job. Agents fire on real events (a letter, a capital notice, a NAV file), produce drafts in the office's own templates, keep every figure tied to its source, and stop before anything posts or reaches the committee. That is more credible for fiduciary work than a chat assistant, and the vertical focus means none of the other finance agents in this directory is a close substitute. The catch is that everything sits behind a demo: no pricing, no public docs, no API, unnamed models and integrations shown only as logos. Treat it as an implementation project. Bring one recurring workflow, ask to see it run on your own documents, and check how exceptions and conflicting values are handled before committing.

About Finpilot

Finpilot calls itself the AI-native operating system for institutional capital, and the product has three layers. Connected Data resolves the office's systems and documents (portfolio systems such as Addepar, Solovis and BlackRock Aladdin, CRMs such as Salesforce and DealCloud, research tools, SharePoint or Box, email and meeting platforms) into one institutional record, matching the same fund or person across sources while keeping each value tied to its document, page and authoritative system and leaving conflicts open for review. Institutional Intelligence answers questions across that record with the evidence, definitions and scope behind each claim. Agents then run recurring work on top of it. Each agent is a configured workflow rather than a prompt: a trigger (a manager letter, capital notice or NAV update arrives, a committee cycle opens, a meeting ends), a preparation step (a cited reporting section, a draft IC book in the institution's template, matched records with exceptions) and a review point where the team decides. Finpilot's examples include reading quarterly letters into the reporting template, checking capital calls, distributions and NAVs against the administrator's file, matching capital notices to commitments before anything posts, flagging manager drift or key-person changes with the passage that shows them, drafting IC memos that recall what the committee said last time, and a meeting notetaker that proposes CRM updates and follow-ups for approval. What sets it apart from general-purpose assistants is that sources, permissions, formats and approval points are configured before the work runs, and Finpilot's team does the data connection and setup, which it says takes a matter of weeks. The tradeoffs are the ones of a sales-led institutional product: no published pricing, no self-serve trial, no documented API, and value that depends on how much of the office's data and process gets connected. It fits allocators whose teams spend their quarters turning PDFs into reports, not individual investors or corporate finance teams.