Pulse, a cutting-edge document processing startup, has announced a $3.9 million seed funding round led by Nat Friedman and Daniel Gross (NFDG), with additional backing from Y Combinator, Sequoia Capital Scout, Soma Capital, Liquid 2 Ventures, Olive Tree Capital, Tiferes Ventures, and executives from NVIDIA, OpenAI, and Ramp.
Founded by Sid Manchkanti (CEO) and Ritvik Pandey (CTO), Pulse addresses a long-standing challenge in the business world: handling critical data trapped in spreadsheets, PDFs, and other complex documents. Despite the presence of PDF parsing and OCR tools for decades, businesses continue to lose substantial portions of valuable information due to inefficient document extraction methods.
READ: Pulse plans to test and integrate Gemini 2.0 for document processing (February 7, 2025)
“Pulse plans to expand our engineering team in 2025 with the new funding,” Manchkanti shared exclusively with The American Bazaar.
Pulse has developed a new approach by combining intelligent schema mapping with fine-tuned extraction models that maintain enterprise-grade accuracy while preserving the context of each document. This innovative platform has already been adopted by Fortune 100 companies, YC startups, and growth-stage enterprises, all benefiting from enhanced retrieval accuracy and significant time savings.
“Our platform is empowering organizations to unlock the full potential of their data,” said Manchkanti in his announcement post earlier today. “We’ve seen firsthand how teams lose up to 30% of critical data due to poor extraction, and Pulse’s solution is closing that gap.”
READ: Inside ClickMe: A conversation with Professor Thomas Serre and Jay Gopal on advancing explainable AI (February 19, 2025)
The team at Pulse, based in San Francisco, is focused on transforming how companies handle data ingestion. Looking to the future, Pulse plans to expand its capabilities to handle multimodal file formats, such as audio and video, to generate even higher quality training data. “Audio and video file formats will allow us to generate higher quality training data,” wrote Manchkanti via email.