Farsight.

It is 11 p.m. and the deck is still not finished.

Somewhere in Midtown, an analyst is on the fourteenth version of a slide that will be glanced at for nine seconds. The logos are misaligned. The footnote font is wrong. The managing director wants the comps re-run by morning. This is the part of finance nobody puts in the recruiting brochure - and it is exactly the part Farsight decided to delete.

Farsight is a New York software company that sells a deceptively simple thing: time. It is an AI system that does the production work of finance - the pitch decks, the models, the buyer lists, the investment memos - directly inside Excel and PowerPoint, formatted to each firm's brand and house style. Not a chatbot you visit. A workhorse that shows up where the work already lives.

The company is three years old, has roughly three dozen people, and just raised $16 million. Its customers are the kind of firms that do not put their vendors on billboards. And its core argument is one every junior banker already believes in their bones: the smartest people in finance spend too much of their lives formatting.

The grunt work that eats the genius.

Here is the inconvenient truth about high finance: the work that creates value - judgment, relationships, the read on a deal - sits on top of a mountain of manual labor. Pull the data. Reconcile the data. Drop it into a template. Re-brand the template. Fix the formatting. Do it again because the numbers moved.

Farsight likes to cite a figure that sounds almost too tidy: 75% of junior finance professionals say they spend up to a quarter of their time on manual tasks. A quarter. The people firms recruited for their brains, spending one day in four being a very expensive copy-paste machine.

The cost is not just hours. It is slower deals, late insights, and the quiet attrition of talented people who did not sign up to be human macros. Everyone in the industry knew this. The clever bit was deciding it did not have to be permanent.

Three from MIT, one who'd lived it.

Farsight was founded in 2022 by three MIT graduates - Samir Dutta, Noah Faro and Kunal Tangri. The split of experience is the whole story. Dutta, now CEO, did time at Evercore and General Atlantic, which is to say he had personally suffered the 11 p.m. deck. Faro and Tangri brought machine-learning depth from Hugging Face and Amazon.

One side knew exactly where the pain was. The other side knew how to build the thing that would actually survive contact with a compliance department. That combination is rarer than it sounds. Plenty of AI startups can demo something dazzling. Very few can ship something a bank's security team will sign off on.

Their bet was contrarian for 2022: that the winning product in finance would not be a smarter chatbot, but an AI that disappears into the existing workflow. No new tab to learn. No "please paste your data here." Just the deliverable, done, in the format the firm already uses.

Not a chatbot. A workhorse.

What Farsight actually does is unglamorous in the best way. You ask, in plain language, for a deliverable. It produces the whole thing - a full pitch deck, a model, a memo - styled to your firm's brand, drawing on your firm's prior work so the output sounds like you, not like a search engine.

It reads unstructured data and works inside native tools: Excel, PowerPoint, PDFs. After acquiring Presentable AI, it can edit and reconfigure slides through natural language - the small miracle of telling a deck what you want instead of dragging text boxes for an hour. And its agents are designed to be proactive, surfacing ideas rather than waiting to be asked.

Then there is the part that wins the actual contract: security. Farsight is SOC-2 compliant, encrypts end to end, lets customers control data retention, and does not train its models on customer data. Firms can run it in Farsight's hosted environment, their own cloud, or fully on-premise. For an industry that treats data leakage as an existential threat, that last sentence is the entire sales pitch.

What you can actually do with it

Build a pitch book for a prospect before lunch. Generate a buyer list without three days in a database. Spin up a valuation model inside the spreadsheet you were already in. Draft an investment memo that reads like your firm wrote it. Edit a fifty-slide deck by typing what you want changed. The common thread: the human keeps the judgment and hands over the typing.

The numbers that made investors lean in.

Talk is abundant in AI. Traction is not. Farsight's case rests on a year, 2024, in which the things that are supposed to grow actually did - revenue up tenfold, customers up fivefold - across investment banks, private equity firms, hedge funds and wealth managers.

The $16M Series A in June 2025 was led by SignalFire, with RRE Ventures, Link Ventures and K5 alongside. The most telling line on the cap table is the angels: operators from Blackstone, Oaktree, Searchlight and Bank of America - people who, by day, are the exact customers Farsight is trying to win. When your buyers invest their own money, the demo probably worked.

The partnerships point the same direction. Integrating S&P Capital IQ Pro and PitchBook means the institutional-grade data lives inside the workflow, not in a separate window someone has to alt-tab to. The Presentable AI acquisition added the slide-editing layer. Each move closes the gap between "ask" and "done."

Automate work. Amplify strategy. Win more.

Farsight's stated mission is to automate financial workflows and decision-making so leaders across finance can work faster and smarter. The three-word version - automate work, amplify strategy, win more - fits on a slide, which is fitting for a company that lives in PowerPoint.

Underneath the slogan is a real position on where AI belongs. Not replacing the banker. Replacing the part of the banker's day that any banker would happily give up. The judgment stays human; the production goes to the machine. It is a modest-sounding promise that, scaled across a firm, changes how fast deals can move.

Back to the 11 p.m. deck.

The skeptic's question is fair: is this a feature or a company? Plenty of tools can draft a slide. The harder thing - the moat, if there is one - is doing it in a firm's exact voice, on its data, under its compliance rules, in the software it already owns. That is less a clever model and more a thousand unglamorous integrations. It is also exactly the kind of thing a bank will pay for and a generic chatbot will never bother to build.

If Farsight is right, the production layer of finance becomes something you describe rather than something you assemble. The associate stops being a formatter and goes back to being an analyst. The work that needed a human all along gets more of the human.

So return to that analyst in Midtown, still awake, still on version fourteen. In Farsight's version of the night, the deck builds itself in the firm's brand, the comps re-run while the coffee brews, and the analyst spends the saved hours on the one thing the software cannot do - having a better idea. That is the bet. Three years in, the people who do this for a living are quietly writing checks on it.

See it move.

Product demos and founder interviews tend to surface on Farsight's own channels. The two best places to look: