AudaciousAgent-native operating systems for modern institutions.
Brain Co. builds end-to-end agentic workflows for enterprises and governments: underwriting and claims in insurance, permitting and licensing in government, prior authorization in health. The company started with proven components, fits them to existing workflows, and deploys in weeks. Many of the products are the first of their kind.
Dan Ashton
CEO. Previously co-founder and CEO of Serene AI (acquired), VP at Clubhouse, and GM and Head of International at Faire. Stanford, where he was student body president, a Board of Trustees member and won the Award of Excellence.
Revant Kapoor
Head of Applied AI/ML. Previously co-founder and CTO of Serene AI (acquired), built and ran the ML team at Clubhouse, and spent nearly five years at Pinterest leading the Trust & Safety ML team. Founding engineer at Highlight, acquired by Pinterest.
Mircea Pașoi
Head of Engineering. Previously co-founder and CTO of Serene AI (acquired), head of engineering at Clubhouse, where he grew the org from ~15 to 40+, and head of identity engineering at Coinbase. Earlier co-founded Memo.AI, acquired by Coinbase.
Brain Co. in their own words.
From coffee to company.
“Why aren’t the biggest institutions implementing cutting-edge AI?” Jared Kushner put that question to Elad Gil over coffee in February 2024, at Elad’s office in Dogpatch.
Kushner and his business partner Luis Videgaray had spent months with leaders of large institutions and kept meeting the same gap: those leaders knew they needed an AI strategy, but had no route to the talent who could deliver one. In Silicon Valley the opposite held, with the best AI talent moving fast and little access to the institutions where the work would matter most. Gil’s answer was to build the company that bridged the two.
As told by Eric Wu, Opendoor co-founder and one of Brain Co.’s incubators
One operating system per institution.
Each product takes the everyday work of an institution, the parts that still run on forms, queues and review, and rebuilds it as an agentic workflow fitted to that organisation’s own data and outcomes.
Permitting, licensing, procurement
The everyday work of local government, fitted to each authority’s data, workflows and outcomes.
Underwriting and claims
Underwriting, claims, authorization, and the operations that sit underneath the policy.
Prior auth, denials, intake
The work between a patient and their care, built for clinical and regulatory constraints.
Reporting and reconciliation
Every number traced back to its source, auditable down to the line item.
The platform underneath
Sovereign-grade security, no model lock-in. Every operating system is built on it.
Engineers from the hardest systems
Built by engineers out of Coinbase, Twitter, Pinterest, Rippling, Anduril and Plaid, across San Francisco, New York, Abu Dhabi, Doha and Jordan.
Recent coverage.
Introducing Brain Co.
What started as a bridge between AI talent and large institutions, expanded into real-world applications. Read the post →
Brain Co. emerges from stealth with a $30M Series A
Incubated by Elad Gil, Jared Kushner, Luis Videgaray and Eric Wu. Read on Yahoo Finance →
The startup helping businesses figure out how to use AI
Forbes on the company and the institutions it works with. Read on Forbes →
My summer as a platform engineering intern
From a first-day metering fix to a company-wide inference cost-attribution platform, shipped to production. Read the post →
What they’re hiring for.
Applied ML Engineer is the focus right now, open across junior, mid and senior, with product and platform engineering alongside it.
Applied Machine Learning Engineer
- Core ML fundamentals, not AI engineering alone. You think rigorously about evals and probabilistic systems.
- A bachelor’s in CS is the floor; an advanced degree is common on the team but not required.
- Open across seniority, roughly a third junior, a third mid, a third senior. New grads have been hired and done well.
- Motivated, takes pride in the work, comfortable in a startup.
Product engineering
Entrepreneurial, customer-adjacent work that continues after the sale is made.
Platform engineering
Backend scalability underneath Atlas and the operating systems built on it.
How the interview runs.
Two screens, then onsite
An ML screen and a coding screen, then an onsite: an ML round, an AI coding round on a representative ML problem where every AI tool is fair game, and a deep-dive on past work with a manager.
Coding, design, then onsite
A coding round and a design round, then an onsite covering coding, design, AI coding and past projects.
If you want ML work with consequences, let’s talk.
Agentic workflows inside insurance, government and healthcare, deployed in weeks and often first of their kind.