About
Making artificial intelligence
as ubiquitous as databases.
Every company runs on databases. Soon, we believe every company will run on artificial intelligence - not as a research project, but as the operating system that handles most of the work. That is what we are building.
- 10,000+customers worldwide
- 34%of Fortune 500 trust Trail
- 1B+agent runs completed
The story
From a sidewalk in Mountain View to the Fortune 500.
Two classmates set out to teach machines to read the world’s paperwork. They are still going.

- 2017Y Combinator, batch W17
- 2022$10M Series A, led by Elevation
- 2024$29M Series B, led by Accel
- 10,000+customers today
Before the sign
Classmates first.
Sarthak Jain and Prathamesh Juvatkar met as electrical engineering students at IIT Gandhinagar. By 2012 they were training models to read muscle signals and hand gestures, years before the deep learning wave.
Their first company, Cubeit, used machine learning to turn web pages into shareable mobile cards. Myntra acquired it in 2016. The exit was the easy part. What stayed with them was harder to shake: inside every business they met, skilled people spent their days retyping documents into software.

The early office was wherever the whiteboard was. The build
A system, not a demo.
The early architecture changed as fast as the field did. Convolutional networks first, then transformers, each model retrained on every new customer's documents to find what held up in production. Accuracy on a benchmark was never the goal. Accuracy on your invoices was.
The company grew the same way: an engineering team in India, a business team in the Bay Area, and a habit of shipping before celebrating.

The Bangalore rooftop, mid-monsoon. Some standups got interrupted. Today
Where the work happens now.
Trail runs agents for finance, healthcare, logistics, manufacturing, and more. More than 10,000 customers use our platform, including companies like Volkswagen, Procter & Gamble, and Ryanair.
Increasingly, the work happens in customers' buildings, not ours. Understanding a reconciliation workflow from the inside is still the fastest way to automate it.

The Palo Alto office.
The founders
Fourteen years, two founders, one unfinished problem.

Sarthak decoded muscle signals at Johns Hopkins and built gesture recognition at HP before most people had heard of deep learning. Prathamesh was a Caltech research fellow and holds IIT Gandhinagar's Institute Gold Medal. They have built companies together since 2012.
The team
The team that ships together.
Automation is the product. Time is the point. Once a year, everyone gets on the same plane.

Boarding call for the Vietnam offsite. 
Ha Long Bay, Vietnam. 
Hanoi backstreets, by jeep. 
Wednesday night football, Bangalore. 
Team dinner, San Francisco. 
Where it started. Mountain View, 2017.
Backed by
Where we are
San Francisco, CA.
- Address
- 2261 Market Street #4010, San Francisco, CA 94114
- info@nanonets.com
- Phone
- +1 650 382 8676
See it run on your context,
with your agents.
Consolidate your company brain and connect your first agent in minutes. Free to start, no card required.
Prefer to read first? See what we build →



