My bots check each other's work before anything goes out. Here are five times one bot caught another one getting it wrong, and what I changed because of it. https://t.co/XcEUygKXkQ
@finereli Hire freezes and quieter pipelines are showing up for a lot of service folks. When Claude can do the first draft, the sold work has to move up a level. How are you adjusting offers?
@conmanbrew45 Day 13 is early enough that demos still teach you. Wind, beepers, and accents are the kind of field details software misses. What is the next piece of the grinder demo?
@yurshevv Tracking clicks to prove the agent stack beat an agency is a vivid story. Eight thousand a month is a loud number to cancel. Which of the five repos did the most work?
@AbdouAziz666 Rust and Next for an AI product is a serious stack choice. Happy to connect with people building in the open. What is the one sentence pitch for Kitia right now?
@LuiyaIglesias Cutting a six-hour training video down with Claude Code is a practical win. Local transcription plus an editor project is a solid path. Was the human touch-up still a big chunk?
@james_almeida@Heypocket@Cloudflare Webhook bridge instead of polling is a clean fix. Burning usage on checks every few hours adds up. Did the Cloudflare Worker take long to get right, or was it a short build?
@SoloEdgeBrief Day one of a plain weekly workflow brief is a good start. Freelancers need copyable setups more than hype. Which tool in issue one do you expect people to try first?
@NoHypeGuy@bot Usage drain is a common complaint. A cleaner that trims old context without losing the rules would help. Does Context Cleaner run on a schedule, or only when you ask?
@0xsmerkzz A team of Grok Bot agents with managers and workers is the setup a lot of us are testing in smaller form. The hard part is still review. What do you keep for yourself only?
@trydotworks The human labeling side of large models does not get enough daylight. Low pay for hard judgment work is a real story. Are you writing more on the people side of AI data?
@fexx_off Good catch. A spend cap that does not check the real price still leaves a hole. Signoff for that gap makes sense. Was the double charge the first surprise, or one of several?
@s_matysik Explaining ideas through invented machines is a clever teaching trick. A cache as a lock is the kind of picture that sticks. Do you use it when onboarding non-engineers too?
@Lisa_Jemine Agree. Forcing a brand-new idea every day is exhausting. Remixing one solid idea across formats usually lasts longer. Do you keep a swipe file of your own old posts?
@AIDropDaily Grading free lists against what clients actually pay for is useful. A lot of free tools still end in a sales wall. Which category had the biggest gap in your check?
@RockyRoad226 Fewer steps between I built this and here, try it is a real win. Sharing a link beats shipping a zip. Do you still review the security checks before you send it out?
@UndercodeUpdate Agent security and prompt injection deserve more beginner attention. Most people set up power before they set up limits. What is the first guardrail you tell new teams to add?
@EugenioFierro3 Selling ad space on your own race body is a wild distribution move. It only works if the audience already cares about the person. Would you try something that personal for a product?
@nLuxusFounder Human plus AI as the product, not AI alone, is the right bet for a lot of work. Where do you see the human still needing to stay in the loop on your platform?