22 comments

  • LogicCraft678 13 minutes ago ago

    I still manually organize my gallery, especially screenshots and downloads

  • rho_soul_kg_m3 an hour ago ago

    Knowledge retrieval. Something where I could ask a question, say in natural language, and I would get an answer. Of course the responses have to make sense and not be made up.

    • RanHal 35 minutes ago ago

      I opened my Obsidian directory with Augment Code, and it indexed it in seconds. Now all my written knowledge is retrievable. You could probably get the same effect with any other RAG-based indexed agent

  • jordiburgos 3 hours ago ago

    Furniture design.

    Give the measures of a bookself and get the pieces and materials needed.

  • SyntaxErrorist 11 hours ago ago

    Tried a few workflow tools but most either break often or need more setup than the task itself. Would easily pay for something that just quietly handled all of that in the background.

    • lishunsheng 8 hours ago ago

      "Curious what kind of workflows you're talking about – are these dev-related tasks or more general business processes?"

      • SyntaxErrorist 2 hours ago ago

        Mostly a mix of both. Some are dev related workflows while others are more operational like content pipelines, research tasks and internal automation stuff.

        • RanHal 32 minutes ago ago

          For dev workflows, Augment works smoothly for me, and I've heard the same about Claude Code and Codex. What specifically breaks?

          The recently announced Cosmos Agent OS is built to handle everything explicitly in the background, but it's more for enterprise than for individuals

  • yen223 5 hours ago ago

    Drawing vector graphics.

    Image generators can make reasonable-looking raster images. LLMs are good at coding. But drawing SVGs sits at the worst of both worlds.

    • RanHal 33 minutes ago ago

      Drawing SVGs via LLMs is mid, but how about converting raster images to SVGs? That sounds like something that shouldn't be too hard

  • late_night_fix 6 hours ago ago

    I still manually synthesize information from multiple sources(docs,blog posts and treads).LLM help with sumaries but merge into a reliable decision step is very human-heavy.

    • judahmeek 4 hours ago ago

      Do you think that can actually be automated? Seems like a judgement-heavy process to me.

  • AiReadyApi an hour ago ago

    Do the dishwasher and throw the rubbish out

  • itsyounish 7 hours ago ago

    I'd say if you could automate product sheet details with detailed image recognition would be really interesting

  • austin-cheney 12 hours ago ago

    I have nothing that fits that criteria exactly. I still write tests for test automation manually and that should be automated but new tests are only needed as features are created or retired, which is irregular and infrequent.

  • pants2 10 hours ago ago

    Folding my laundry

    • codegeek 4 hours ago ago

      Some companies are building Robots for this already. for example https://figure.ai

      • _aavaa_ 2 hours ago ago

        Except that robot probably costs more than the prevent value of all clothes I’ll ever buy.

        • codegeek an hour ago ago

          I hear you. I also don't buy clothes unless really really needed. But I am hoping that if I do buy a Robot like that, it will do other things as well and not just folding laundry :)

    • aimiraclemag 8 hours ago ago

      Ahahaha, good point! Some automation on that field would be precious.

  • fieldsate 6 hours ago ago

    preparing for a trip , travel, lots of moving pices like itenary , packing , tickets , IDs , cabs .no sync .huge mess

    • jp42 2 hours ago ago

      I believe this will get solved once agent protocol is adopted widespread.