Finbar Is Already Inside the Hedge Funds You've Heard Of

The Macro: Wall Street’s AI Problem Is Not What You Think

Every financial services company says they use AI now. It’s in the annual reports, the investor decks, the recruiting pitches. What most of them actually mean is that someone on the data team fine-tuned a model to tag earnings call transcripts, or they bought a Bloomberg terminal add-on that highlights sentiment. The real analytical work, the financial modeling, the comp table construction, the deep research on a company’s unit economics, is still done by junior analysts pulling 80-hour weeks in Excel.

This is a $40+ billion market for financial data and analytics, dominated by Bloomberg, Refinitiv, FactSet, and S&P Capital IQ. These companies sell data access and basic tools. The analysis itself remains stubbornly manual. A senior portfolio manager asks a question. An analyst spends two days building a model to answer it. The model gets reviewed, revised, and presented. The decision gets made. By the time the analysis is finished, the market has already moved.

The AI coding tools like Cursor and Replit aren’t solving this because financial modeling isn’t a general coding problem. It requires domain-specific knowledge about how to structure a DCF, what adjustments to make for different industries, how to read footnotes in SEC filings, and how to reconcile data from multiple sources that don’t agree with each other. General-purpose AI writes code. Financial AI needs to write the right code for a specific analytical framework.

There’s a growing cohort of startups trying to wedge into this space. AlphaSense does AI-powered research. Hebbia targets document analysis for finance. Kensho (S&P Global) handles analytics on structured data. But the full-stack AI analyst that can go from question to financial model to recommendation, replacing the two-day analyst workflow with something that takes minutes, is still mostly vapor across the industry.

Finbar says they’ve already built it, and that the customers are already real.

The Micro: Financial Models in Minutes, Not Days

Finbar is an AI investment analyst that automates financial modeling, research, and data analysis. The product is used by several top-20 hedge funds measured by assets under management, which means the companies using it collectively manage hundreds of billions of dollars. That’s not a soft metric. Hedge funds don’t adopt tools for fun. They adopt tools that make them money or save them enough time that it amounts to the same thing.

The platform works through a web-based dashboard and an Excel add-in. The Excel integration is a smart move because it meets analysts where they already work. Telling a financial analyst to abandon Excel is like telling a surgeon to abandon the operating room. It doesn’t matter how good your alternative is. The workflow habits are decades deep. Building into Excel instead of replacing it shows product judgment.

The core capabilities cover three areas. First, automated financial modeling: building and updating models in a browser or Excel within minutes instead of the hours or days a human analyst would need. Second, AI-powered research across earnings calls, filings, and market data. Third, access to global financial data spanning public and private companies. The combination means an analyst can go from “what does this company’s free cash flow look like under different revenue assumptions” to a completed, editable model without starting from scratch.

Testimonials from equity portfolio managers at top hedge funds report significant time savings on routine analytical tasks. The company claims a 100x reduction in time spent on manual data work, which is a big number, but financial modeling genuinely is that manual. An analyst building a three-statement model from 10-K filings can easily spend eight hours on data entry alone before any actual analysis happens.

Founder and CEO Edward Huang was previously a hedge fund analyst at Balyasny and Goldman Sachs and is a CFA charterholder. He wrote most of the MVP himself, which means the product was built by someone who has actually sat in the seat, pulled the data, built the models, and presented the analysis. That’s worth a lot in a domain where understanding the workflow details is the difference between a useful tool and a useless one. Co-founder and CTO Robin Gan was a software engineer at Thought Machine, working on cloud-native banking platforms. The team is four people, and they came through YC W25.

The Verdict

Finbar has the most credible customer proof of any company I’ve covered in this space. “Several top-20 hedge funds” is specific enough to be verifiable and impressive enough to matter. Hedge funds are among the most demanding, security-conscious, and results-oriented customers in any industry. If Finbar is surviving inside those environments, the product works.

The competitive position is strong but not permanent. Bloomberg could build this. FactSet could build this. AlphaSense is moving in the same direction from a research-first angle. The hedge fund world is also full of internal quant teams that build proprietary tools. Finbar’s advantage right now is speed and focus. They’re a small team building exactly one thing for a market they understand from the inside. The incumbents are large companies with extensive product lines and slower development cycles.

What I’d push on is the scalability beyond hedge funds. The buy-side is a great starting market because the customers are rich and demanding. But it’s also a relatively small market in terms of total seats. The real business question is whether Finbar can expand to asset managers, equity research shops, investment banks, and eventually corporate finance teams. Each of those markets has different workflow requirements, and what works for a PM at Citadel might not work for an FP&A analyst at a Fortune 500.

The founding team’s pedigree, the customer list, and the product approach are all strong. This is a company that shipped before it pitched, and the people using it are the ones who would know if it doesn’t work. That’s as close to product-market fit validation as you get in enterprise software.