Today in AI ·

Ai2 #OpenSource #AIInfrastructure #AIResearch

Ai2 releases Olmo-core 3 for large mixture-of-experts training

Ai2 released Olmo-core 3, an open training framework redesigned for mixture-of-experts models and benchmarked beyond one trillion parameters. It keeps experts resident on GPUs and routes data to them, reducing repeated weight transfers in large training runs.

Read at Ai2
Anthropic #ClaudeCode #DeveloperTools #AI

Anthropic launches TypeScript mods for Claude Code

Anthropic introduced mods that let developers change Claude Code prompts, tool behavior and interface elements through TypeScript functions packaged as plugins. They work in the CLI and desktop app and run with Claude Code’s machine access, making trusted distribution important.

Read at Anthropic
Google #Google #Accessibility #Gemini

Google launches Guided Vision in Gemini Live for Android

Google launched Guided Vision in Gemini Live on compatible Android devices, offering camera-based audio descriptions and verbal reframing cues developed with blind and low-vision communities. It expands conversational visual assistance but is not intended for navigation or obstacle detection.

Read at Google
Nature #AIResearch #ReinforcementLearning #Evaluation

Nature study shows Ataraxos mastering hidden-information games

A new Nature paper presents Ataraxos, which defeated a leading Stratego player while using substantially less compute and data than previous efforts. Its self-play and search techniques also performed strongly in other hidden-information games, broadening evidence for efficient decision-making methods.

Read at Nature
Bilibili #OpenSource #Translation #AIModels

Bilibili releases Index-Translate models for 150 languages

Bilibili’s Index Team released the Index-Translate family, including an Apache-2.0-licensed 35B-total, 3B-active mixture-of-experts preview supporting text translation across 150 languages. The model follows terminology, formatting and style instructions, offering an open option for controlled translation.

Read at Bilibili
Anthropic #EnterpriseAI #Banking #Claude

Barclays expands Claude across its global operations

Barclays expanded its collaboration with Anthropic to apply Claude to software development, legacy-system modernization and operational efficiency. The bank expects Claude Code adoption to reach half its developers by the end of 2026, signaling a broader rollout in a regulated industry.

Read at Anthropic
DoorDash #AIAgents #Commerce #DoorDash

DoorDash opens beta waitlist for AI ordering by text

DoorDash opened a US beta waitlist for a text-based ordering assistant that finds food, suggests a cart and handles checkout within the conversation. The launch brings its AI shopping experience beyond the app and can remember preferences for future orders.

Read at DoorDash
Anthropic #Government #EnterpriseAI #Claude

Claude for Government becomes generally available

Anthropic made Claude for Government generally available to federal and state agencies through a FedRAMP High authorized environment. Claude Code CLI and Claude for Microsoft 365 are entering early access within the same environment, extending governed AI access for public-sector work.

Read at Anthropic

Free · 18 episodes · no sign-up

Learn AI system design, one diagram at a time

60-second Shorts and long-form deep dives on how real AI systems are built, how they break, and how to fix them: caching, RAG, agents, security and decision models.

Newest episodes

All 18 episodes

The learning path

Five stages, from the edge of the network to the security of your agents.

Open the learning path

0 of 18 episodes done

Mark episodes as done from their page. Progress stays in this browser.

Start with stage 1
  1. 1 Foundations Edge & Caching Why far servers feel fast, where edge rendering backfires, and how one cache expiry can take down an AI app. 3 episodes · 3 coming 0 of 3 done
  2. 2 Core LLM Apps: Caching & RAG Put retrieval and caching in front of a model without leaking data or serving the wrong answer. 2 episodes · 2 coming 0 of 2 done
  3. 3 Core Decision Models When a typed decision beats generated text, and how to route between a decision model and an LLM. 3 episodes · 1 coming 0 of 3 done
  4. 4 Advanced Agents in Production Context, memory, cost guardrails and the workspace an autonomous coding agent needs. 8 episodes · 6 coming 0 of 8 done
  5. 5 Advanced AI Security Stolen tokens, poisoned tools and the other ways an AI system gets turned against its users. 2 episodes · 2 coming 0 of 2 done

Four ways to watch

Get the next episode first

Shorts and deep dives on AI system design land on @AI.JoinDev first. Subscribe, or follow the RSS feed.

Teaching someone younger?

Code Quest Kids: Coding for kids aged 6 to 12, with Bit the robot. Same team, same no-ads, no-tracking rules.

Visit kids.join.dev