River AI raises $1.1B to reinvent AI agents and training
Igor Babuschkin’s River AI bags $1.1B in two-month-old startup funding...
River AI, the two-month-old AI startup founded by former xAI and DeepMind researcher Igor Babuschkin, has raised $1.1 billion in a seed/Series A round led by General Catalyst and AMP PBC, with participation from Nvidia, AMD Ventures, Y Combinator, and Temasek. The company emerged from stealth in June with a bold mission: rebuilding AI from the ground up, starting with training methodologies, to create “personally trainable” agents rather than replacements for human workers.
Babuschkin argues that the current AI trajectory prioritizes scale and labor displacement, whereas River aims to deliver agents that act as “guardian angels”—quietly integrated into daily life, deeply personalized, and owned by the user. The startup’s initial offering is an API that supports reinforcement learning (RL) and low-rank adaptation (LoRA) fine-tuning, billed per 1 million tokens, with enterprise-friendly features like 15–20 minute RL runs and 2–4x cost savings versus closed-source alternatives. Its long-term vision includes hardware co-design to enable local, personalized AI agents.
- River AI, founded by ex-xAI/Igor Babuschkin, raised $1.1B led by General Catalyst with Nvidia, AMD, YC, and Temasek participating.
- API supports RL/LoRA fine-tuning, charges per 1M tokens, and promises 2–4x cost savings with 15–20min enterprise RL runs.
- Goal: build end-to-end AI systems for personal, trainable agents that act as always-on “guardian angels,” not human replacements.
Why It Matters
Could redefine AI ownership and personalization—shifting power from closed labs to individuals and enterprises.