SAN FRANCISCO: Baseten, a company powering inference for some of the world’s fastest-growing artificial intelligence products, said on Thursday it has acquired Parsed, a reinforcement learning startup specializing in post-training and continual learning for large models.
The deal brings Parsed’s team of researchers and engineers into Baseten, unifying inference, data, evaluation and post-training in a single platform. Financial terms of the acquisition were not disclosed.
“Inference is the stickiest and most important part of the AI stack,” Baseten co-founder and CEO Tuhin Srivastava said. “What makes inference truly valuable over time is continual learning: using real production data and evaluations to train better, faster, cheaper models.”
Parsed, founded by Oxford PhD researchers and Rhodes Scholars Mudith Jayasekara and Max Kirkby along with Chief Scientist Charles O’Neill, has worked with Baseten customers as a partner prior to the acquisition. The team’s expertise in reinforcement learning and post-training is expected to strengthen Baseten’s ability to deliver specialized, high-performance models.
Jayasekara, Parsed’s CEO, said the combination would accelerate the shift away from general-purpose models toward specialized systems. “We believe the world doesn’t need ever-bigger general-purpose models, it needs models that deeply understand a specific job and keep getting better at it,” he said.
The acquisition follows Baseten’s $150 million Series D fundraising round, which valued the company at $2.15 billion and brought its total capital raised to more than $285 million.