AI VIDEOS
One-sentence takeaways and concise summaries of important AI videos.
How to Test Your Agents on Real Tasks With Harbor - Xander Song, Arize AI
Xander Song explains how realistic tasks, isolated environments and outcome verifiers can help teams test agent changes, then demonstrates a Phoenix experiment that reduces unnecessary tool calls.
Make $$$ Selling AI to Boring Businesses
Greg Isenberg and James Dickerson explain how AI-assisted research, landing pages and follow-up can support local service businesses, emphasizing booked jobs and measured results over selling agents as a novelty.
OpenAI Cuts Model Price by 80%
Jack Roberts discusses reported model price cuts, a specialist image model comparison and resistance to AI-assisted software, weighing cost, reliability and local access.
How I’m Actually Using Claude Duet
Nathaniel Whittemore explains where personal agents fit into his content workflow, finding more value in selected proactive tasks than in replacing every existing AI interaction.
Humanoid Harassment: Do Robots Deserve Rights?
Brent A. Anders’s discussion of robot rights separates speculative consciousness from present-day human responsibility, using alleged incidents to question how consent, liability and accountability should be handled.
One Captain Runs a Whole Crew of AI Coders
Ray Fernando compares long-running coding-agent orchestrators, describing a central captain that delegates work across cloud and local machines and contrasting visual interfaces with terminal tools.
How to Build an AI Code Review Agent - Daksh Gupta, Greptile
Daksh Gupta explains how targeted prompts, incident-based evaluation, codebase context and runtime tests can improve an AI review agent while keeping cost and false-positive feedback under control.
Build Your Own Team of Managed Agents - Jack Wotherspoon, Google DeepMind
Jack Wotherspoon explains cloud-managed agents and demonstrates separating conversation state from a shared sandbox so specialized agents can pass files and work between stages.
These 3 skills changed my life
Theo Browne explains how he turns repeated coding workflows into skills, separating bounded review, authorized automatic merging and preparation that waits for human approval.
How to Build Voice Agents That Don't Fall Apart When You Interrupt - Jesse Hall, LiveKit
Jesse Hall argues that reliable voice agents need conversation-level evaluation, audible latency measurements and asynchronous tools, not isolated model rankings.
Every AI app on the iPhone Duo ranked
Alex Finn compares four AI apps on the iPhone Duo by their use of the larger display, sidebar navigation, artifacts and access to coding sessions.
Master ChatGPT Work in Under 20 Minutes
Alex Kerss demonstrates how persistent project instructions, reusable skills, connected apps and scheduled work can turn a one-off AI task into a repeatable workflow.
How to Write Your First AI Eval - SallyAnn DeLucia, Arize AI
Sally-Ann DeLucia demonstrates a practical evaluation loop: inspect real agent traces, define concrete pass and fail criteria, test one quality and compare the judge with human feedback.
What Does an AI Engineer Do Now? - Aparna Dhinakaran, Arize AI
Aparna Dhinakaran argues that AI engineers increasingly build the systems that evaluate, monitor and improve agents, rather than only implementing individual features.
The AI Boom Has a Profit Problem
Jack Roberts argues that rising AI usage is not the same as measurable business value, then considers how automation changes creative work, learning and agent payments.
Opus 5.5 Is Proof That We Do Need Smarter Models
Theo Browne argues that stronger coding models unlock useful software creation for technically curious people as well as professional developers.
Stop Paying for AI, Run Everything Locally (Here's How)
David Ondrej and Ahmad M. Osman argue that local AI needs optimized hardware, useful tools and a smooth experience, with cloud models reserved for selected tasks.
How Monzo Built a Coding Agent on Existing Platform Tools
Felippe Felisola Caso and Francesco Galletta explain how Monzo combined existing platform capabilities with a sandboxed coding agent to support engineering and business workflows.
MiniMax M3.1 Flash, Model Costs and AI Trust
Nick Saraev and Jack Roberts examine cheaper AI models and argue that benchmark scores, deployment controls and user trust require separate scrutiny.
Competing Visions for Governing Advanced AI
TheAIGRID examines competing ideas for governing advanced AI, distinguishing reported scenarios, international oversight proposals and speculative first-mover control from established capability or policy.