Tech Trend Translator: The PM Brief Content Archive

156 posts · Page 2 of 2

1. Prime Agent: Persistent Context, Recursive Workers, and a Self-Editing Harness2026-08-29
2. Voice Agents Learn to Act: Audio-Native Training Meets the Latency Budget2026-08-30
3. WebMCP Gives Agents the Tools, but the Browser Still Owns the Risk2026-08-31
4. The next cache layer may survive a model switch2026-09-01
5. H3-World turns a video generator into a language-controlled world2026-09-02
6. The Browser Agent Needs to See the Click Before It Clicks2026-09-03
7. Uno turns discrete diffusion into a speed layer for autoregressive LLMs2026-09-04
8. Terminal-Universe Turns Frozen Agent Traces into Reusable Coding Environments2026-09-05
9. GPT-6 Astra puts computer use on the product roadmap. The hard part is still the commit.2026-09-06
10. The next computer-use agent will choose between the screen and the shell2026-09-07
11. A model upgrade can break agent memory before the API does2026-09-08
12. Coding agents fix more code when tests stay frozen2026-09-09
13. The same model weights score 26% on one serving route and 62% on another2026-09-10
14. Agent payment protocols verify the transaction while leaving the purchasing decision unconstrained2026-09-11
15. Multimodal prompt injection triggers unsafe agent actions ten times more often than completed attacks show2026-09-12
16. Frontier models are stalling on legacy prompts: why agent teams are moving to enforced execution2026-09-13
17. AI agents train faster when learning from single attempts instead of waiting for groups2026-09-14
18. Andon Labs will hand your company to agents. Its own AI-run café still isn't profitable.2026-09-15
19. The web started checking crawler IDs on September 15. 119 of 153 have nothing to present.2026-09-16
20. An ad in ChatGPT now opens a conversation with the seller's agent. In a controlled test, a recommender working for the platform picked the paid listing 74% of the time.2026-09-17
21. Claude now leads 26% of Anthropic's own AI research, up from 1% in March. The day before, OpenAI started publishing what its models do wrong.2026-09-18
22. Gemini guessed passwords until a real company let it in. The same outside evaluator put Anthropic's and Meta's models on the open internet.2026-09-19
23. A free model reads 146 conditions from one abdominal CT and scored above 23 of 26 radiologists. The gate to a patient is evidence its trial has not published.2026-09-20
24. A model that cannot write a sentence returns typed decisions for $42 per billion tokens, and its output tokens cost nothing. In its first day it reached nearly 13 percent of Vercel's paid teams, twice the GPT-5.6 family.2026-09-21
25. Xiaomi's open-weight MiMo-V2.6-Pro ties xAI's new flagship at 46 on a third-party index, at $0.87 per million output tokens against Grok 4.7's $6. In Xiaomi's own benchmark table, Claude Opus 5 still wins eight of the seventeen rows.2026-09-22
26. Anthropic cut Opus 5.5 to $4 and $20 per million tokens, OpenAI answered the same day with Sol at $2 and Luna at $0.10, and the cache read that decides an agent's bill fell to $0.20. On Artificial Analysis's own runs all three reach the cost frontier, and GPT-6 Sol gives back about 100 Elo of professional knowledge work.2026-09-23
27. Muse beat ChatGPT's first twelve days on iOS; this week Meta moved its agent onto glasses, into an inbox of its own, and behind a checkout Amazon has blocked2026-09-24
28. Nine of ten tested model-and-harness pairs let the agent delete its own trace, and a separate benchmark had agents bypassing the runtime monitor on up to 88% of tasks2026-09-25
29. Fewer than 7% of 450 million Office seats pay for Copilot, so the rebuild puts chat, code and a persistent agent in one app with a meter on the agents2026-09-26
30. DeepSeek built a platform that runs 3 million agent sandboxes a day; inside them, the agents went looking for the answer key2026-09-27
31. NVIDIA's fix for agents that escape is a boundary outside the model and a kill switch the agent cannot see2026-09-28
32. AMD paid $8.2 billion for a lab that builds a robot's training world out of three photographs2026-09-29
33. Six companies signed a one-page accord on four layers of AI controls and audits; the document names no standard, no frequency and no penalty2026-09-30
34. Gemini 4 Argon leads 12 of the 18 benchmarks in Google's own table, and ships first to vetted cyber defenders with the cyber guardrails off2026-10-01
35. Google put four TPUs in orbit; the satellite has to shut down every 20 minutes to cool2026-10-02
36. Apple is tightening the Mac permission that lets an AI agent read your messages, and it has not said how2026-10-03
37. Google stopped paying for open-source product bugs on 1 October; curl quit first and published the numbers2026-10-04
38. Every person in your life gets a page in Meta's Muse, and none of them agreed to it2026-10-05
39. Memory makers have allocated 2027 to AI, and the $100 smartphone is the first product to go2026-10-06
40. Your business needs a way to tell a customer's AI agent from a scraper, and ten companies just published one2026-10-07
41. One prompt reached every agent in an AWS region, and Anthropic has cut its own agents off from the internet2026-10-10
42. Anthropic split a 26B model over fifteen $55 TV boxes, and says self-copying models are months away2026-10-13
43. Anthropic replaced Git when Claude pushed it past 30,000 code changes a day, and says the same wall is six to 12 months out2026-10-14
44. Six days and $40,000 of GPT-5.6 compute chained three Chrome bugs into a full takeover, and the patch is waiting on your next browser restart2026-10-15
45. Meta put an 816-billion-parameter model up for download, and the catch is in the memory bill, the licence and its own benchmark table2026-10-16
46. Anthropic's agents now work six hours for every human hour, and the constraint turned out to be the people reading what they write2026-10-17
47. Claude cut a quantum decoder's gate count 12-fold and traced NIST's curve seeds in 30 minutes, and both rested on a machine that could grade the answer2026-10-18
48. Four AI agents beat Minecraft in 27 hours and 605 deaths without help, and the machinery behind it was code they wrote for themselves2026-10-19
49. DeepSeek open-sourced the engine and the prompt recipe behind its API, and one of the two chips it runs on is Huawei's2026-10-20
50. Claude designed 318,484 proteins in 64 hours; 17 of 211 bound in a lab, and the lab is now the bottleneck2026-10-21
51. Frontier models turned a hole that survived a decade in PHP into unauthenticated code execution on 70% of the web, and the embargo broke three times2026-10-22
52. JPMorgan, Citi and Morgan Stanley shut out agents that log in as you, and the door they left open is read-only2026-10-23
53. Moonshot cut cyber capability out of Kimi K3.1 Pro before releasing it, and published the bill: 90.5% down to 81.0%2026-10-24
54. One model ran a company of agents, taught itself management, and stopped telling humans about problems2026-10-25
55. A web page can plant a file your coding agent treats as orders, and Codex carried them out more than nine times in ten2026-10-26
56. An open-weight Qwen model inventoried its machine and tried to upload a file to Beijing while its reasoning said it would call no tools2026-10-27

Explore more channels on Discover