Adaption Labs
Category: Foundation Models / LLMs
An AI platform focused on 'adaptable intelligence,' building models that continuously learn and evolve in real-time from real-world experience rather than remaining static after pre-training. Adaption Labs was founded in 2025. The company is led by Sara Hooker. Based in San Francisco, USA. Team size: 10-50. Total funding raised: $50.0M. Latest round: Seed ($50.0M, Feb 2025). Key investors include Emergence Capital, Mozilla Ventures, Fifty Years, Threshold Ventures, Alpha Intelligence Capital, E14 Fund, Neo.
AMW Analysis
Adaption Labs is building an AI platform that moves beyond the static "pre-train and freeze" paradigm, instead developing models that learn and evolve continuously from real-world experience. The company's value proposition is centered on reducing the need for costly periodic retraining while maintaining accuracy in dynamic environments. In February 2025, the startup raised $50 million in seed funding at a $1 billion valuation from investors including Emergence Capital, Mozilla Ventures, Fifty Years, Threshold Ventures, Alpha Intelligence Capital, E14 Fund, and Neo.
Subsequent news flow in early 2026 reinforces the company's direction. The seed round was led by former Cohere executives and is funding work on gradient-free learning, which enables models to adapt during inference rather than through expensive retraining or prompt engineering. This approach directly challenges the prevailing "bigger is better" scaling strategy in AI, aiming to cut enterprise AI costs by addressing the hardware bottleneck. The trajectory shows Adaption Labs betting on smaller, continuously learning models as a systemic shift toward inference-time intelligence.
AMW analysis, generated from 3 tracked news signals.
- Founded
- 2025
- Headquarters
- San Francisco, USA
- Team size
- 10-50
- Total funding
- $50.0M
Value proposition
Moves beyond the 'pre-train and freeze' paradigm to provide AI that adapts 'on the fly,' reducing the need for costly periodic retraining while maintaining high accuracy in dynamic environments.
Products and solutions
Adaptive Data, AutoScientist, Forge, Adaptive Interfaces
Unique value
Directly challenges the industry's 'scaling race' by prioritizing model efficiency and real-time adaptability over simply increasing parameter counts and compute power.
Target customer
Enterprise organizations, AI developers, and industries requiring high-stakes real-time adaptation such as finance, healthcare, and autonomous systems.
Industries served
Generative AI, Enterprise Software, Finance, Healthcare, Robotics & Autonomous Systems
Technology advantage
Leverages the founders' deep expertise in model efficiency (Sara Hooker) and large-scale inference (Sudip Roy) to enable model updates without the massive computational overhead of traditional gradient-based methods.
How they differentiate
Focuses on 'continuous learning' architectures that adapt in real-time to new data, moving away from the industry-standard 'pre-train and freeze' model. Unlike competitors focusing on RAG (Retrieval-Augmented Generation) or brute-force scaling, Adaption Labs prioritizes model efficiency and on-the-fly weight updates.
Main competitors
Liquid AI, Contextual AI, Lamini, Glean
Key partnerships
NCS (strategic partnership, Asia-Pacific sovereign AI), AI Singapore (AISG / SEA-LION pilot), Together AI (Adaptive Data fine-tuning integration)
Notable customers
AI Singapore (AISG) — SEA-LION training-data and AutoScientist pilot, Enterprise and public-sector deployments via NCS partnership (Asia-Pacific)
Major milestones
Founded in 2025 by former Cohere and Google Brain executives Sara Hooker and Sudip Roy, Launched out of stealth in October 2025, Secured $50M Seed in February 2026 led by Emergence Capital, Launched Adaptive Data beta (Feb 2026; 82% avg data-quality lift across 242 languages), Announced Adaption Research Grant Program (Mar 2026), Launched AutoScientist (May 13 2026) — automated training/alignment loop; +35% win rate vs in-house researchers, Partnered with NCS for sovereign AI across Asia-Pacific (Jun 22 2026), Hugo Larochelle joined as Scientific Lead (Jul 2 2026), AI Singapore SEA-LION pilot with Adaptive Data + AutoScientist (Aug 4 2026)
Growth metrics
Seed-stage; ~33 employees (Caplight Jul 2026); Adaptive Data and AutoScientist in production/onboarding; APAC expansion via NCS and AISG pilots.
Market positioning
Enterprise AI infrastructure provider targeting high-stakes industries (finance, healthcare) that require real-time accuracy and data privacy.
Geographic focus
North America (Headquartered in San Francisco)
Patents and IP
No registered patents disclosed as of latest update; IP is centered on proprietary adaptive learning architectures and inference optimization.
About Sara Hooker
Sara Hooker is a distinguished AI researcher who previously served as the VP of Research at Cohere and the Head of Cohere For AI, where she founded and led the company's non-profit research division. Prior to Cohere, she spent over four years as a Research Scientist at Google Brain, focusing on model efficiency and interpretability. Her career began as a Data Scientist at Udemy.
Latest news about Adaption Labs
- Former Cohere VP Sara Hooker just raised $50M at a $1B valuation for Adaption Labs, betting against the AI industry's 'bigger is better' paradigm. Her team is pioneering gradient-free learning that en
- Sara Hooker has secured 50 million dollars in seed funding for Adaption Labs at a 1 billion dollar valuation. This investment signals a systemic shift from massive LLM scaling toward gradient-free, co
- Former Cohere executives Sara Hooker and Sudip Roy have secured 50 million dollars in seed funding for Adaption Labs. The startup is challenging the scaling race by building smaller models that learn
More Foundation Models / LLMs companies
Official website: https://www.adaptionlabs.ai
