Breaking
$33M Series A led by Prosus Ventures Shortcut scores 80%+ on Excel World Championship cases Backed by Patrick Collison and Eric Schmidt Fairies: a general-purpose AI assistant ships Robert Yang trades MIT faculty post for SF agent lab 35 people. 4 teams. One thesis: digital humans. $33M Series A led by Prosus Ventures Shortcut scores 80%+ on Excel World Championship cases Backed by Patrick Collison and Eric Schmidt Fairies: a general-purpose AI assistant ships Robert Yang trades MIT faculty post for SF agent lab 35 people. 4 teams. One thesis: digital humans.
Vol. I - Profile No. 042 San Francisco, CA Series A - 2025
The Lab Dispatch

Fundamental
Research Labs

A 35-person studio in San Francisco teaching software to behave less like a tool and more like a colleague. The polite name for that ambition is "digital human beings."

Fundamental Research Labs - company brand image
Frame 01 The lab calls them digital humans. Their first one knows Excel cold and bills by the seat.

A quiet office, a loud spreadsheet.

On a Tuesday afternoon in San Francisco, an analyst opens a blank Excel workbook. She types one sentence. Forty seconds later, the workbook contains a discounted cash flow model, sensitivity tables, and a sources tab. She did not write a single formula. The thing that did is called Shortcut, and it is the most visible product of a 35-person studio called Fundamental Research Labs.

The lab is small enough to fit in one room and ambitious enough to use a phrase that should embarrass it - "digital human beings" - without flinching. The team has decided that the next chapter of software is less about chat boxes and more about agents: programs that can plan, act, and keep going after you close the laptop.

Founder Robert Yang spent his earlier career as a computational neuroscientist on the MIT faculty, which is a useful clue. His lab is not chasing the biggest model. It is chasing the most useful one - the one that can carry a long task across hours, tools, and people without losing the plot. The bet is that "socially intelligent" agents are not a feature; they are the product.

We are building digital human beings - autonomous, collaborative, and socially intelligent agents.- Fundamental Research Labs, company brief

There is a games team. There is a prosumer apps team. There is a core research team and a platform team. From the outside it looks like a media company crossed with a research institute. That is roughly the idea. Products are not just things to sell. They are instruments. Each one is a way to measure what an agent can actually do in the wild.

Shortcut, the Excel agent, is the loudest instrument so far. When Fundamental ran a blind study, managers from McKinsey and Goldman Sachs preferred Shortcut's work to first-year analysts from their own firms in 89% of comparisons. On Excel World Championship cases - the closest thing the spreadsheet world has to a varsity sport - it scored over 80% and finished in roughly one-tenth the time. The video clips of one-prompt DCF models went viral on LinkedIn the way recipes used to go viral on Pinterest.

Then there is Fairies, the general-purpose consumer assistant. The pitch is older than the iPhone: an assistant that knows your apps, answers across them, schedules things, takes action. The graveyard of attempts is wide. Fundamental's argument is that the model layer has finally caught up to the dream, and that the missing ingredient was never raw intelligence but social context - the ability to behave like a person who has met you before.

In August 2025 the lab announced a $33 million Series A led by Prosus Ventures, with Patrick Collison participating. That brought total funding to roughly $45.1 million. Earlier checks came from a16z Speedrun, Patron, First Spark Ventures, Factorial, VamosVentures, and Eric Schmidt. The investor list is short, opinionated, and tells you something: this is being treated as a research bet that happens to ship.

Productivity apps are where the most value is being created today - but eventually we want to solve physical problems.- Dr. Robert Yang, founder

Yang has been candid about the longer arc. Productivity is the entry point because it pays now. The destination is embodiment - agents that work alongside people in the physical world. That is the part that makes you understand why a games team exists alongside an Excel team.

The lab used to be known as Altera. The old domain, altera.al, still serves as a leftover email. The rename to Fundamental Research Labs was not cosmetic. It announced a shift away from one product and toward a portfolio - and toward the more uncomfortable claim that what they are building is, in some way, human-shaped.

Whether or not you buy the framing, the products are doing the talking. The lab is charging users for Shortcut after a seven-day trial and is reportedly already revenue-generating. Fairies is in the wild. The hiring page is busy. And the analyst at the start of this story is now on her sixth model of the day.

$33M
Series A (Aug 2025)
89%
Blind-test win rate vs. MBB analysts
80%+
Excel World Championship score
~10x
Faster than human on those cases
35
People on the team

Two products. One thesis.

Fundamental Research Labs ships under several brands. Each one is, in effect, a research probe wearing a marketing budget.

Prosumer - Finance

Shortcut

An Excel agent that converts a natural-language prompt into a full model - DCFs, sensitivities, simulations - with traceable steps. Subscription-based after a 7-day trial.

Consumer - Assistant

Fairies

A general-purpose AI assistant that chats, plugs into your apps, queries your knowledge bases, and books things on your calendar. The pitch you have heard for a decade, finally with the model layer to back it.

Internal - Research

Agent Platform

The shared substrate underneath everything: agentic reasoning, multi-agent coordination, tool use, memory. The thing the games team and the apps team both depend on.

Shortcut, by the numbers

Excel WC score
82%
Blind-test wins
89%
Speed vs human
~10x
Trial length
7 days

A neuroscientist with a shipping habit.

RY
Founder & CEO

Dr. Robert Yang

Yang's background is computational neuroscience, and he ran a faculty lab at MIT before moving full-time into building. That trajectory shapes the company: less "biggest model wins," more "what would a brain actually do here." He has been public about a long arc that runs from productivity software today to physically embodied agents later. The team around him pulls from MIT EECS, Stanford NLP Group, Google X, and Citadel.

A short, fast timeline.

2023
Company founded in San Francisco, originally as Altera.
2024
Closes $9M seed co-led by First Spark Ventures and Patron, with a16z Speedrun and Eric Schmidt participating.
July 2025
Launches Shortcut publicly. Demo videos of one-prompt DCF models go viral on LinkedIn and X.
August 2025
Announces $33M Series A led by Prosus Ventures, with Patrick Collison participating. Total funding crosses $45M.
2025 - 2026
Expands portfolio under the Fundamental Research Labs brand: Shortcut, Fairies, and an internal research platform.

Who's writing the checks.

Series A - Lead

Prosus Ventures

Anchored the $33M round in August 2025 - the strategic capital signal that pulled the rest of the round together.

Seed + Series A

a16z Speedrun

Returning backer across rounds.

Seed Co-Lead

Patron & First Spark

Co-led the seed; both participated again in the Series A.

Angels

Collison + Schmidt

Patrick Collison (Stripe) and Eric Schmidt (ex-Google CEO) wrote angel checks across the rounds.

What's actually shipped.

Demos & conversations.

Public videos and walkthroughs from the team and the press. (External links - open in a new tab.)

Back to the spreadsheet.

The analyst from the opening scene closes her laptop at 6:14 p.m. The models are done. Six of them. She did the thinking; Shortcut did the typing. The strange part is not that the workbook exists. The strange part is that she will do it again tomorrow, and the day after, and quietly forget what the old way felt like. That is what Fundamental Research Labs is really shipping - not a feature, but a default. A room of researchers in San Francisco has decided the next decade of work will be done with a colleague who does not need lunch. Whether they are right is a story their products keep answering, one prompt at a time.

Pass it on.