Trent Conley

@SpaceX AI

[Trent Conley @TrentJConley]

@cwamidon Robotic farming is quite un-fundable; the majority of farming costs are capital costs not labor.

[Trent Conley @TrentJConley]

@matt_elevenlabs 11B

[Trent Conley @TrentJConley]

@Nicolascole77 L take… inference costs will dramatically reduce over time. Take a look at the same cost per unit of intelligence one year ago. It’s decreasing at roughly 10x. This is not even taking into account improvements gpus or running AI on specialized hardware

[Trent Conley @TrentJConley]

@arrington Based Arrington

[Trent Conley @TrentJConley]

@Michael_Druggan @grok She could’ve misspoke, intentionally mislead the public to obfuscate Bryan’s identity (they weren’t public yet) or mixed up timelines with a different partner.

Either way, there’s no proof if he cheated or not and claims that he did/did not are useless gossip

[Trent Conley @TrentJConley]

@ADoricko No, socialism still sucks.

Housing is very far from a free market. If you removed zoning, permitting, and allowed people to build you wouldn’t see PE firms buying single family homes because they’d be decreasing in value due to new developments dropping rent

[Trent Conley @TrentJConley]

@eleventhsavi0r @arithmoquine The quality of writing is

[Trent Conley @TrentJConley]

@scottastevenson Good players and high APM players overlap at the highest levels (grandmaster, higher master) but for lower master and diamond they aren’t necessarily the same. You can get by with ~100 APM in diamond/low master.

[Trent Conley @TrentJConley]

I like the intent of the analogy, but in reality macro is more important: both in life and StarCraft

[Scott Stevenson @scottastevenson]

It's surprising how few people play real life like it's Starcraft. APM matters a lot. If you play life this way you will do well, because almost no one else is.

[Trent Conley @TrentJConley]

@signulll Disagree. If they can nail highest level of intelligence with low cost (which seems like they did, beating sonnet) then they’ll win the api market which will imo be much larger than consumer facing chat bot market

[Trent Conley @TrentJConley]

Major props to @cursor_ai team with the 1.2 release. Give it a try, you can ask the agent questions about a codebase and it’ll generate diagrams that are accurate, among many other agentic improvements

[Trent Conley @TrentJConley]

@karpathy Use silicon earplugs. They take a week or two to get used to then your sleep quality will be great even in noisy environments

[Trent Conley @TrentJConley]

@Erdayastronaut Same here Tim

[Trent Conley @TrentJConley]

@ns123abc Prove P=NP

[Trent Conley @TrentJConley]

@measure_plan tutorial!

[Trent Conley @TrentJConley]

@iamgingertrash That’s a hot take. They’re currently winning in AI or at least extremely competitive, still winning in search, winning in content (YouTube) and winning in self driving cars (to be seen how it hold up with Tesla, but there’s no robo taxi yet, just waymo).

[Trent Conley @TrentJConley]

@cursor_ai Way better than sonnet. Has slightly higher intelligence but doesn’t go off the rails editing things in agent mode

[Trent Conley @TrentJConley]

@getomni_ai Nice benchmark, matched my internal testing showing Gemini 2.0 flash still the best overall.

[Trent Conley @TrentJConley]

@emollick If you build a Docker image with llama on it and have that in a CI/CD pipeline on GitHub push actions that’ll probably count to their 1 billion but add a lot more than actual uses in production