The wiki

A materials-discovery company integrating AI, atomistic simulation, and robotic laboratories to design and test new materials.

Chemist and materials researcher combining machine learning, physical simulation and experimental data, and Lila Sciences’ chief scientific officer for physical sciences.

Retrieval-augmented generation connects a language model to external information. Its central engineering problem is choosing evidence that helps the model answer—and checking that the answer actually uses it.

LlamaIndex’s early evolution shows why connecting language models to data requires more than a larger context window.

Rahul Sonwalkar, known publicly as Rahul Ligma, founded Julius AI to help people use code for data analysis without having to write it themselves.

Railway is a cloud deployment platform expanding from application deployment into agent-driven app building and release control, supported by its own compute infrastructure.

Part 2 of the Railway conversation follows self-deploying workloads, durable workflows and cloneable environments to a practical question: how does faster code generation become safe production software?

Railway’s closing ambition is an instantaneous path from idea to shareable production deployment, pursued at every necessary layer of the stack.

Deploying an application is only the beginning. Part 1 follows Railway’s approach to evolving it through environment versioning, owned hardware, agent interfaces and safer production experiments.

Product leader behind Google’s NotebookLM and cofounder of Huxe, focused on turning source material into useful, personal AI experiences.

Data scientist associated with Auditoria AI who advocates automating finance-agent repair workflows while preserving human verification and accountability.

Caylent’s chief technology officer writes about bringing generative AI into production, with an emphasis on evaluations, user workflows, economics and the authority of agents that take action.

Humanloop cofounder and former CEO whose work spans probabilistic machine learning, expert-guided model development and evaluation of language-model applications.

Andon Labs tests whether agents can sustain a business, coordinate with people and act reliably in the physical world. Earnings matter—but so do the decisions that produce them.

Gray Swan connects adversarial testing, specialized guard models and system permissions to make agents safer without stripping away their usefulness.

AI product work at Zapier: connecting language models to app actions, designing background workflows, and teaching colleagues to build with AI.

An AI research and product company building multimodal models, enterprise agents, and systems for physical-world perception and generation.

An AI model deployment platform built around open-source model packaging and hosted inference, acquired by Cloudflare in December 2025.

How a tool for sharing research became a platform for running and customizing models—and why packaging, GPU economics and hacker culture all mattered.

A browser development environment that expanded from AI code completion and its own code models into agents that build, test, deploy, and coordinate work across applications and other artifacts.

From financial text analysis to Ghostwriter: a reading edition of the May 2023 conversation with Reza Shabani on training code models, evaluating their feel, and moving beyond autocomplete.

Faster code generation leaves verification work behind. Retool’s bet is on AI inside business processes, with developers controlling the steps and people reviewing the results.

An unaffordable recommendation feature led to Turbopuffer’s storage architecture. Agents issuing many searches at once are now changing its workload and economics.

Databricks cofounder and chief architect whose work spans Spark analytics, database execution and live operational data for AI agents.

Replit’s Head of AI in 2023, Shabani helped build the data infrastructure and team behind Ghostwriter and the company’s own code language models.

Ride Home Media built daily news podcasts around concise subject briefings; its flagship Techmeme Ride Home combined AI coverage with Brian McCullough’s use of AI production tools.

Prompt-engineering practitioner known for public language-model experiments, model comparisons and collaborative research on multi-turn jailbreaks.

Machine-learning researcher working on reinforcement learning, language-model distillation and self-correction, with earlier contributions to reliable RL evaluation.

Cloudflare product leader whose public work on AI agents addresses durable execution, permissions, and the infrastructure needed to delegate tasks safely.

MiraOmics co-founder and CTO working on AI for biological data analysis, and co-host of Latent Space’s AI for Science podcast.