About
Confluence Labs is an AI research lab building systems that design effective experiments and learn efficiently from limited data. It targets researchers and engineers in data-sparse, experiment-intensive fields such as hardware, biology, and materials science; its differentiation is program-synthesis-driven AI models, including a reported 97.9% ARC-AGI-2 score.
Market
Confluence Labs competes at the intersection of frontier AI reasoning, AI-for-science, and autonomous or data-efficient experimentation. Its positioning is to help researchers generate highly informative hypotheses and learn from limited experimental data, differentiating through LLM-based program synthesis, discrete search, long-horizon refinement, and code-execution verification rather than only scaling a general-purpose model; its closest overlaps are broad AI-scientist systems such as Google DeepMind and FutureHouse and domain-focused materials-discovery companies such as CuspAI and Orbital Materials.
Confluence Labs is aimed at research-intensive organizations—particularly hardware and semiconductor engineering, biotechnology and drug-design, materials science, and physics teams—where scientists and engineers must make decisions from sparse data and expensive experiments. The likely buyers and collaborators are technical researchers, domain experts, and industrial R&D leaders; the company has not disclosed a specific company-size segment or mature commercial-sales model.
At a Glance
Problem
Confluence Labs targets a central weakness of modern AI: models perform well when abundant training data is available, but struggle in scientific domains where data is sparse, expensive, or slow to collect. Designing new molecules, discovering physics, and engineering advanced materials are constrained search problems in which every physical experiment costs time and money; the killer use case is helping researchers identify the most informative next experiment so that breakthroughs in areas such as drug discovery, materials science, fusion, and climate technology require fewer trials.
Product / Service
The company describes itself as an AI research lab building models that learn efficiently from experience. Its intended system generates testable hypotheses and designs effective experiments in data-sparse domains, then extracts as much value as possible from each result. The technical approach combines large language models, program synthesis, discrete search, and structured long-horizon work so models can generate, evaluate, and refine candidate solutions. Public materials emphasize a research-and-partnership model rather than a clearly packaged commercial SaaS product, with the company seeking collaborators in biology, materials science, hardware engineering, and other fields where physical experimentation is the bottleneck.
Market
Confluence Labs sits at the intersection of AI for science, scientific discovery, active learning, and AI-driven experimental design. Its adjacent competitors include FutureHouse, whose agents automate literature review, hypothesis generation, experiment planning, and data-driven discovery, and Atinary, whose SDLabs platform recommends experiments, learns from results, and optimizes complex laboratory processes. A third-party company database also lists Nota, Fiddler Labs, and Portkey as competitors, although the precise overlap with Confluence Labs’ scientific-reasoning focus is unclear.
The company is very early stage rather than a proven commercial vendor. It was founded in 2025, entered Y Combinator’s Winter 2026 batch, has a two-person team, and has publicly reported $500,000 in Y Combinator funding. Its clearest traction is technical: a reported 97.9% score on the ARC-AGI-2 benchmark at approximately $12 per task, with the solver open-sourced. Public materials do not identify paying customers or revenue, so Confluence Labs is best characterized as pre-commercial or still validating how its core learning-efficiency technology will be applied to scientific R&D.
Founders & Leadership
Funding History
Y Combinator
Recent News
StartupHub.ai profiled Confluence Labs after its 97.9% ARC-AGI-2 result. The article explains that 12 LLM agents generate and refine Python transformations in parallel sandboxes, and highlights intended applications in drug design, hardware engineering, and physics research.
Imbue’s ARC-AGI-2 analysis referenced Confluence Labs’ latest high score and described its approach as using Gemini CLI agents working in parallel. The coverage places Confluence’s result in the broader evolution of ARC-AGI-2-solving systems.
Skyfall.ai discussed Confluence Labs’ 97.9% ARC-AGI-2 approach, which uses several Gemini-CLI instances—REPL agents with command-line access—working in parallel.
Confluence Labs announced that it was coming out of stealth with a state-of-the-art ARC-AGI-2 score of 97.9% at approximately $12 per task on the public evaluation. The company also said it had open-sourced its solver and described a program-synthesis approach driven by LLMs.
Confluence Labs’ official launch post introduced its AI research lab and reported the 97.9% ARC-AGI-2 result at roughly $12 per task. It linked to the open-source solver and said the lab was developing partnerships with researchers and engineers in areas including hardware, biology, and materials science.
Y Combinator lists Confluence Labs as an active San Francisco company in its Winter 2026 batch, founded by Brent Burdick and Niranjan Baskaran. Confluence Labs’ website states that it is backed by Y Combinator and Paul Graham; no funding amount was reported in the available evidence.
Active Roles
0No active roles right now.
Get notified when they postBusiness Model
The company has no publicly disclosed pricing or customer revenue model. Available reporting indicates it had no paying customers, revenue, API documentation, or pricing at the time of research; it is backed by Y Combinator and Paul Graham while developing its technology and research partnerships.