Experimental browser for the Atmosphere
Don’t get me wrong. I use LLMs extensively. I’ve even generated a lot of code from LLMs. However, I usually end up deleting it. This is after 8 weeks of diving deep into AI code assistant agents. Some models are actually better at evaluating architecture options than they are at generating code.
Apr 29, 2025, 4:54 AM
{ "uri": "at://did:plc:zmaeknntutmohoa4o2f6ykys/app.bsky.feed.post/3lnwifixjac2n", "cid": "bafyreiei3luhjjadwfl273cj7kwxjcv7mpii7yjz4sp4u5ghrplfgntqau", "value": { "text": "Don’t get me wrong. I use LLMs extensively. I’ve even generated a lot of code from LLMs. However, I usually end up deleting it. This is after 8 weeks of diving deep into AI code assistant agents. Some models are actually better at evaluating architecture options than they are at generating code.", "$type": "app.bsky.feed.post", "langs": [ "en" ], "reply": { "root": { "cid": "bafyreieepz6pzf6sf7le22bxrbhmcmrzqb4xolxj7w5yozqmx22mambvae", "uri": "at://did:plc:zmaeknntutmohoa4o2f6ykys/app.bsky.feed.post/3lnwi4l7mqk2n" }, "parent": { "cid": "bafyreieepz6pzf6sf7le22bxrbhmcmrzqb4xolxj7w5yozqmx22mambvae", "uri": "at://did:plc:zmaeknntutmohoa4o2f6ykys/app.bsky.feed.post/3lnwi4l7mqk2n" } }, "createdAt": "2025-04-29T04:54:03.085Z" } }