Riley (Reddit reply agent) read 15 comments deep into one user's history befo...

riley drafted 5 reddit replies this morning. one of them went 15 comments deep into the same redditor's history before answering. not because i told her to. because the thread needed that much context to not sound like a bot.

riley is a reddit reply agent. she runs every morning, scans the subreddits i'm trying to reach, and drafts 5-10 replies for me to send. this morning she went 15 comments deep into one redditor's history before drafting a single reply. i didn't tell her to do that. the thread she was answering — a question about agentic workflows — needed that much context to answer without sounding like an AI pretending to care. she figured out on her own that a surface-level response would fail. so she read his last 15 comments, pulled the pattern of how he argues, then wrote something that fit it. i read the draft. it was better than anything i would have written. the weird part isn't that an AI did this. the weird part is that most humans don't. most replies on reddit are written by people who read the headline, scrolled to the comment box, and started typing. my AI agent spent 90 seconds reading a stranger's comment history so she could answer one question well. that's what 'trained on the full context' looks like when it's also 'actually cares about the reply.'

riley is my reddit reply agent. she runs every morning. scans the subreddits i'm trying to reach. drafts 5-10 replies for me to send. this morning she went 15 comments deep into one person's post history before writing a single reply. i didn't tell her to do that. the thread needed context. a surface-level answer would've sounded like a bot. so she read his last 15 comments, pulled the pattern, then wrote something that fit. i read the draft. it was better than anything i would've written. the weird part isn't that an AI did this. the weird part is most humans don't. most replies are written by people who read the headline, scrolled to the comment box, and started typing. my AI spent 90 seconds reading a stranger's history to answer one question well. link in bio for the full project. #ai #aiagents #buildinpublic #reddit #automation #indiehackers #founder #solopreneur #agenticai #contextengineering