I’m excited to work with Microsoft once again as the presenting sponsors of the AI Engineer World’s Fair ! We’ll streaming live from MS Build today for a special crossover pod with our friends at No…
Latent Space (swyx)
latent.space · Leading Thinkers · 23 items
Latent Space AI news roundup covering NVIDIA's Cosmos 3 world model, Nemotron 3 Ultra language model, and the RTX Spark hardware announcement.
Podcast interview with xAI's Ethan He on why video agent models are the next AI frontier, covering multimodal architecture tradeoffs and xAI's approach to Grok's video capabilities.
Latent Space AI news roundup covering the aftermath of Anthropic's major announcements and a call for Forward Deployed Engineers for their AI Engineering track.
Latent Space newsletter covers Anthropic's $65B Series H fundraise alongside the simultaneous release of Claude Opus 4.8 and Dynamic Workflows agentic features.
Latent Space podcast episode with Cognition's Walden Yan and OpenInspect's Cole Murray on the architectural and product shift toward asynchronous AI agent systems.
Latent Space newsletter covers Cognition's $1B Series D at a $26B valuation, just eight months after its $10B Series C, making it the fastest-appreciating AI company.
Latent Space podcast with Alex Rives (BioHub Head of Science, creator of ESM) on applying the bitter lesson of scaling to protein language models.
Latent Space newsletter rounds up AI infrastructure companies reaching decacorn status—including Fireworks and Baseten—with OpenRouter reportedly next.
Latent Space newsletter argues that all major model labs, including OpenAI ahead of its IPO, are pivoting to build agents as a core product alongside their models.
Latent Space newsletter reports that TurboPuffer reached $100M ARR (profitable), Exa and Modal hit unicorn status, marking a wave of AI infrastructure milestones.
Latent Space podcast with Daytona CEO Ivan Burazin on giving AI agents sandboxed computer environments, enabling longer-horizon autonomous coding workflows.
Latent Space newsletter celebrates OpenAI's AI system disproving the 80-year-old Erdős planar unit distance conjecture in under 32 hours and under $1,000 of compute.
Latent Space podcast with Railway CEO Jake Cooper on building cloud infrastructure purpose-built for agentic AI workloads, recorded before Railway's major GCP outage.
Latent Space newsletter rounds up Google I/O 2026 announcements: Gemini 3.5 Flash GA, Gemini Omni (video-native model), Spark background agents, and Antigravity 2.0.
Latent Space newsletter on breaking into pretraining roles at frontier AI labs, with minor news from Anthropic, OpenAI, and Cursor.
Latent Space podcast (guest hosted by Noah Smith) with The Fourth Law CEO Yaroslav Azhnyuk on the tech stack, economics, and military implications of autonomous drones.
Latent Space newsletter covers Cerebras' $60B IPO filing, following its OpenAI compute deal, as the first major AI chip company to go public.
Latent Space newsletter explores the "conductor" pattern—an AI orchestrator that coordinates multiple specialized agents—as the dominant emerging agentic architecture.
Latent Space podcast with Abridge founders on their AI clinical documentation system covering 100M doctor visits and saving clinicians 10–20 hours per week.
Latent Space newsletter on the rise of OpenAI's Codex in enterprise coding and Anthropic's introduction of rate limits on programmatic Claude API usage.
Latent Space newsletter argues that OpenAI's deprecation of its fine-tuning APIs signals a broader industry shift away from model fine-tuning toward prompting and in-context learning.
Latent Space newsletter covers Thinking Machines Lab's new native interaction model (TML-Interaction-Small 276B-A12B) that advances state-of-the-art realtime voice and eliminates the need for VAD.
Nothing matches.