Episode publish date
October 7, 2025 5:00 PM (UTC)
Last edit date
Oct 21, 2025 4:57 PM
Last snip date
October 21, 2025 4:27 PM (GMT+1)
Last sync date
October 21, 2025 5:56 PM (GMT+1)
Show
Last Week in AI
Show notes link
Snips
20
Warning
⚠️ Any content within the episode information, snip blocks might be updated or overwritten by Snipd in a future sync. Add your edits or additional notes outside these blocks to keep them safe.
‣
Your snips
‣
[04:54] Text-To-Video Crossing Into Photorealism
‣
[11:40] Agent-Focused Model Improvements
‣
[18:10] Structure Agent Loops With Verification
‣
[22:27] Feed Framing Can Create 'AI Slop' Backlash
‣
[26:45] Use Connectors Carefully For Daily Briefs
‣
[33:47] Safety Routing Acknowledges Emotional Risks
‣
[36:04] Small Models Power Scalable Agentic Features
‣
[40:02] Prefer Native Agents In Productivity Apps
‣
[42:54] Agentic Commerce As A Revenue Lever
‣
[46:09] Thinking Machines' 'Tinker' Fine-Tuning Tool
‣
[50:08] Liquidity Spike Raises OpenAI Valuation
‣
[55:48] AI + Robotics To Automate Science
‣
[58:37] Benchmark Long-Horizon Software Tasks Properly
‣
[01:01:33] Concepts Emerge In Two Training Phases
‣
[01:05:45] Concise Chain-Of-Thought Is Often Better
‣
[01:12:16] Short Attention Windows Aid Long Memory
‣
[01:18:18] California's Frontier AI Transparency Law
‣
[01:21:31] Adopt Proven Provenance Standards Early
‣
[01:25:31] Character.ai Removes Disney Characters
‣