The wiki

Chroma cofounder whose work spans embedding-based retrieval, training-image attribution and long-context language-model evaluation.

Zed co-founder whose work spans editable AI conversations, open edit prediction, and the editor’s collaborative and performance foundations.

The company behind Cursor developed custom coding models and reinforcement-learning agents alongside its editor, expanded toward code review, and announced a completed SpaceX acquisition in August 2026.

Biomedical machine-learning researcher and builder of exploratory interfaces for large language models.

The company behind iPhone, iPad, and Mac combines generative AI products, custom silicon, and tools for local machine learning.

A physical AI company that expanded from autonomy-development tools into vehicle operating systems, learned driving systems, and agentic engineering workflows.

Perplexity co-founder whose work combines cited AI answers, search engineering and a focus on product quality.

AI researcher whose collaborative work spans direct preference optimization, evaluation of AI feedback, and autonomous robot learning.

Real user tasks make Arena’s evaluations valuable. Funding those requests, retaining users, and deciding what to publish make the benchmark a product and governance challenge.

AI researcher and DatologyAI co-founder whose work spans neural-network representations, generalization, and training-data curation.

OpenAI computer-use leader and co-founder of Sky, whose work connects mobile automation with AI agents that operate software.

Tenex co-founder and software engineer who advocates paying engineers for accepted output, with incentives to use AI effectively and safeguards for code quality.

Cube co-founder and CEO whose work connects governed semantic definitions with AI-driven analytics and human control.

An AI benchmarking company comparing model capabilities, inference performance, and cost, with research extending into factual reliability and agentic work.

Choosing a model means comparing capability, behavior, speed and the cost of completing the work. Artificial Analysis builds evaluations around those trade-offs.

Glean co-founder and CEO focused on company context, enterprise retrieval, and the cost and reliability of AI at work.

Microsoft executive whose AI-platform work emphasized agent orchestration, governance, and feedback-driven model improvement, before she became Xbox CEO in 2026.

Reinforcement learning researcher whose contributions include Genie's latent action model, Gato's early Atari prototype, and imitation learning from observations.

Foundation Capital general partner and enterprise AI investor who, with Jaya Gupta, developed the context-graph thesis: preserving how organizations make decisions as a foundation for AI agents.

Reinforcement-learning researcher whose work spans Berkeley robotics, OpenAI reasoning models and model development at Cursor.

Devin and OpenInspect show what background coding requires: shared sessions, working environments, useful tests, scoped memory and review that keeps repositories maintainable.

Known publicly as Auriel W, Wright has worked on reinforcement learning at Gemini and writes about inspecting model behavior, designing rewards, and choosing economically relevant training tasks.

An AI-agent project created by Toran Bruce Richards that evolved from a standalone autonomous agent into a platform for building, running, and managing automated workflows.

How onboard autonomy changes drone operation—and why mission verification, manufacturing, and the economics of defense matter as much as the AI.

Andon Labs cofounder and Vending-Bench coauthor studying whether autonomous AI agents can sustain coherent decisions and run businesses.

An AI mathematics company developing formal theorem provers and Lean infrastructure, with a stated expansion into provably correct code, cryptography, and quantitative finance.

Formal proofs give AI a checkable target for learning and generation. Axiom’s challenge extends from finding proofs to specifying the right problem, managing large searches, and making results reusable.

An open-source model-training framework that grew from YAML-configured fine-tuning into reinforcement learning, multimodal training and distributed post-training.

Hungarian technologist and Agent Community founder working on AI-agent identity, discovery, and machine-readable publishing.

AI researcher and technology leader whose work includes Jamba’s hybrid language models and a case for input-specific, resource-aware AI execution.