EmbeddingGemma 2 for Local Multimodal Search
Daniel Walsh demonstrates offline multimodal search with EmbeddingGemma 2 and explains where retrieval, media interpretation and query-score calibration remain separate problems.
Your #1 AI video source
AI videos, nothing else.
One-sentence takeaways and concise summaries of important AI videos.
Our Latest AI Videos
Voice Delegation and Daily Workflows with ChatGPT Dots
Riley Brown demonstrates how voice conversations with ChatGPT Dots can delegate coding work, handle communications and produce documents, with scheduled briefings and selective alerts.
Building Custom Clients on the Codex App Server
Dominik Kundel shows how the Codex app server lets developers build custom interfaces around an existing agent harness, while retaining its tools, configuration and event stream.
AI Research Progress and the Trade-offs of Wider Access
AI Copium contrasts reported progress in AI-assisted research with misuse concerns, arguing that optimism about scientific gains should not remove scrutiny of wider access.
Videos in our catalogue by publication date
Codifying Agent Coding Rules with Linters and Policies
Greg Pstrucha argues that repeated corrections to AI-generated code should become repository checks and review policies, rather than instructions that disappear after context compaction.
Practical ChatGPT and Claude Interface Updates
Igor Pogany demonstrates recent ChatGPT and Claude interface changes, distinguishing useful everyday workflows from features that still behave inconsistently in his tests.
The Best New AI Products and the Key Trends They Represent
Nathaniel Whittemore argues that recent AI releases point toward greater model choice, cheaper specialized inference and interfaces generated for the task at hand.
Google Stitch 2.0: The FREE AI Design Powerhouse by Google is CRAZY!
AICodeKing walks through a Google Stitch design-to-code workflow, emphasizing targeted revisions, reusable design rules and checks against the intended user journey.
OpenAI's "Alien Math" Is Freaking People Out
Wes Roth examines the tension between reported AI mathematical progress, human understanding of complex proofs and uncertainty about downstream security consequences.
Theo Browne finds Haiku 5.5 useful for inexpensive, bounded agent subtasks, while his hands-on failures show why low token prices do not guarantee low total task costs.
OpenClaw Upgrades: Team Threads, Tools and Memory
OpenClaw's walkthrough highlights new model options and workflow features, including team threads, pinned tools, memory skills and meeting support.
Why 80% Reliability Isn't Good Enough - Felipe Blanes, Amazon AGI Lab
Agent evaluations should evolve from real customer needs and production failures, rather than treating a static benchmark score as proof of trustworthiness.
How a Remote Company Builds AI Fluency - Em Shreve, Automattic
Automattic combines protected learning time, real projects and peer teaching to help a distributed workforce adopt AI rather than merely distributing tools.
Brains vs Hands: How to Run AI Agents Safely in Production - Viren Baraiya
An agent can propose a plan while a deterministic harness owns execution, durability and required gates, making long-running production workflows easier to control and recover.
We Mapped 115 Microservices for Our Coding Agents - Kamalakannan Nandagopal, Postman
A shared API context graph can help coding agents understand distributed systems, provided every relationship is grounded in code or telemetry and kept current as services change.
Google's Bid for Work Emails: What AI Agents Can Learn
Nate B Jones argues that workplace archives can teach AI agents to imitate activity without capturing the judgement and outcomes that make knowledge work valuable.
Turning Agent Memory Into Reliable Skills - Will Lyon
Will Lyon shows how connected memory and structured reasoning traces can become reusable agent skills whose steps remain grounded and auditable.
AI Model Releases, Creative Tools and Research Caveats
TheAIGRID surveys new models, agent interfaces and creative tools while discussing the limits of research benchmarks and AI-consciousness claims.
Giving AI Agents Production Context - Matt Gibiec
Matt Gibiec argues that agents need live system context, dependency awareness and observable outcomes to make useful changes beyond a local coding loop.
Eight AI Repositories and Skills for Practical Business Work
Greg Isenberg explores eight AI repositories and skills through practical business examples, emphasising small trials, customer evidence and human judgement.