The wiki

MCP’s first year exposed the hard parts of shared agent infrastructure: enterprise identity, bidirectional transport, trusted discovery, long-running work, and neutral governance.

From PyTorch’s hardware tradeoffs to shared feedback and home robots: a reading edition of the March 2024 conversation with Soumith Chintala.

Open models gave builders more control. Shared training pipelines let researchers reuse one another’s work—but compute, data access and commercial incentives still limited participation.

The organization behind ChatGPT, the OpenAI API, and Codex has evolved from nonprofit research into large-scale model development, commercial deployment, and agent platforms.

Part 1 follows developers from simpler conversation APIs and custom GPTs to tool execution, expressive speech, and the unresolved engineering questions behind vision.

Part 2 of Latent Space’s DevDay recording explores conversational programming, managed retrieval’s limits, plugin migration, and Zapier’s move from chat actions to scheduled assistants.

OpenAI’s developer-relations function connects developers with API updates, model releases and practical education about building with AI.

An internal OpenAI research team whose collaborations include SWE-bench Verified and GDPval, with work spanning coding capabilities, professional tasks, and research automation.

Collaborating OpenAI teams behind SWE-bench Verified and GDPval, combining expert human review with evaluations of coding and economically valuable work.

A June 2023 developer roundtable on native function calling, cheaper inference, and the decisions application code still needs to control.

The March 2025 launch conversation explains why Responses became a new API, how its tools compose, and how handoffs and traces could support better agents.

Peter Steinberger’s personal AI agent project grew from a messaging relay into an open platform, with a nonprofit foundation, paid maintainers, and a security-focused ecosystem.

A developer platform that evolved from fine-tuning smaller language models to reinforcement learning for agents, acquired by CoreWeave in 2025.

OpenRouter combines a common API for AI models with routing, failover, billing and enterprise controls. Its platform separates choosing a model from choosing the provider that serves it.

Alex Atallah and Anjney Midha on model distribution, inference competition, fusion, and the security challenges of a growing token economy.

AI researcher whose collaborative work spans sequence-to-sequence learning, neural-network distillation and AlphaStar.

Anj Midha’s case for a compute grid begins with idle accelerators and expands into a theory of frontier labs: pool resources, reduce coordination losses, and let the mission determine what to build.

Microsoft AI Knowledge leader focused on information understanding, retrieval and the infrastructure that grounds AI agents in enterprise data.

Part 3 moves from writing before the experiments are settled to the environments that shape research judgment, then closes with an argument for building AI ecosystems around experienced technical contributors.

Generative-image researcher, Black Forest Labs cofounder, and coauthor of work on VQGAN, latent diffusion and rectified-flow transformers.

Paul Bakaus works on AI-assisted interface design through Impeccable, with an emphasis on human judgment and a shared design vocabulary.

Founder and CEO of Browserbase, building browser infrastructure and automation tools that let AI agents work through existing websites.

As of July 2026, Cursor’s VP of Forward Deployed Engineering, building an enterprise AI deployment team around production experience, organizational adoption, and long-running agents.

Mistral audio research lead on streaming speech generation, enterprise voice personalization, and learning from customer-specific evaluations.

AI researcher and joint SAM 3 project lead, with contributions to Llama visual grounding and VinVL's visual representations.

Materials discovery company combining AI, simulation and physical experiments.

Liam Fedus and Ekin Doğuş Çubuk explain how uncertain measurements, simulation and lab automation fit into materials discovery.

Applied Intuition’s co-founder and CTO develops software for intelligent machines, with an emphasis on operating systems, simulation, and the constraints of deploying AI onto physical hardware.

Creator of OpenClaw who works on agents at OpenAI and emphasizes execution, verifiable feedback and human judgment.

Software developer and author at Baseten whose writing and teaching explain how generative AI models become production services.