... person walks away, because the action did not change, the world did. Which makes "ask again" the part I would push on. Re-asking on every change is expensive, and expensive gates die. People route around them or rubber stamp them. Re-checking is cheap. Declare what the approval depended on, have the runtime re-evaluate it at execute time, and only surface the difference when something actually moved. Yes, provided nothing changed, verified by the system rather than ... ... second question looks identical to the first, the second answer is identical too, and you have converted an accumulation problem into a fatigue one. Somebody tapping yes for the fortieth time is not a control, they are a rubber

... capability and start being about control: work stalls without anyone noticing, it's unclear which agent owns a given output, a "cheap" agent turns out expensive because its work keeps getting redone, and approvals either swarm you or get rubber-stamped into meaninglessness. Those aren't intelligence problems. They're the exact problems a normal human org solves with roles, budgets, an escalation path, and a manager who owns the result. So the bet we're making with ClawTeams ... ... each time. Money and anything touching production are the other two, but the byline one is the one I'd never soften, because a bad deploy is revertible and a bad sentence in my voice isn't. On manager vs

... signing" - that's also the moment of maximum pressure to just get it done. the client's already mentally committed, so a deliverables/kill-fee field that first shows up on the signing screen is more likely to get rubber-stamped than actually read. I'd bet the fields still need to be seen once before that moment, even informally, so the signature is locking something the client already reacted to rather than something they're seeing for the first time under ... ... pure information questions go away, but there ll be a category of can you do X that no form covers. Comment from adam albo(@suchback): @galdayan  You re right that first exposure at the signing screen gets rubber

... devs I know got good at coding by writing a lot of it. But we never really trained to read it the same way. Code review culture is getting worse, not better, as AI output volume increases. People are rubber-stamping PRs they don't fully understand. The bugs that make it to production aren't from bad writers, they're from missed reads. Curious if others are feeling this. Do you actively practice reading code, or is it just ...

... part of this! Comment from Abdullah Javaid(@abdullah_javaid3): Welcome, and good timing to be asking before Saturday rather than after. C2PA signing is the third thing in your list and I think it is the only one a competitor cannot copy by Q4. Fifteen models behind one subscription is an integration sprint. Rolling credits is a pricing decision. Signed provenance is a commitment you hold on every render forever, which is why almost nobody offers it. The question ... ... more honest answer to give heading into Saturday than I could have given two days ago. Comment from Abhineet Arora(@abhineetarora): @abdullah_javaid3  You hit the nail on the head this is exactly where the rubber

... proofs stay up to date without creating a large maintenance burden? Comment from Raj Nagulapalle(@rnagulapalle): @omri_ben_shoham1  it's me, personally .. I read what the adversarial model surfaces and decide what is a real class vs noise. no second automated layer watching the red-team for staleness. and that's intentional.., not just small-team reality (though it's that too): an automated checker on the red-team would just push the calibration problem up a level ... ... problem is keeping the red-team prompts calibrated as the underlying behavior shifts, so right now it's mandatory but with a manual review step to make sure the adversarial model is still asking hard questions and not just rubber

... setup. I still don't have a clean answer for where "let it run" should end and "ask me first" should begin. Genuinely curious how you all handle it. Comment from Gal Dayan(@galdayan): @aadilghani  same read-only-vs-reversible line for me. the thing that changed it wasn't a rule I wrote, it was switching to a tool that asks per-action instead of per-session - so "let it rip" doesn't mean blanket auto-approve ... ... overrode a stable version. Luckily I'm not live yet, or it would've been a big issue. Comment from Stacy Wycoff(@stacywycof83995): @aadilghani  I don't run coding agents but I have the same read-only vs

... solve creative velocity for small brands" would have been "you can't affordably." Now the answer is "AI generation of variations that are actually shippable, not slop." Same problem, different physics. So the split I'd draw is idea vs approach. If AI hands you a problem you haven't personally seen, the risk is you're building for a hypothesis nobody has validated. But if you already saw the problem and AI just unlocks a viable solution, that ... ... hitting personally before I ever opened a chat window. Where AI actually changes things for me is speed of rejecting bad versions of a good idea. I can sketch a rough direction, have it poke holes or sketch three alternate

... whether a behavioral floor, even an imperfect one, is better than no floor at all. Curious if anyone here has built or worked with review quality measurement systems and what signals you found actually correlated with genuine engagement vs. nominal approval. Project: https://www.lineagelens.dev/ Comment from Fabrizio Pfannl(@fabriziowexare): Interesting framing, the 1-sec-per-line floor reads sensible. The more durable signal you might already be sitting on is content-of-comment: did the reviewer reference something ... ... diff itself (a test name, a related ticket, a prior bug)? Bots and rubber

... part I see skipped most. Scoped access and audit trails get talked about, but ownership a specific human who's accountable for what the agent does on their behalf is what makes the approval step meaningful instead of a rubber stamp. And it's striking how the whole thread keeps landing in the same place: design it from day one. Retrofitting after the first incident means unwinding agents already in production and users already trained to expect speed exactly ... ... Elias put it, architecture beats retrofit. The bar moved; most mental models just haven't caught up yet. Reading-vs

... Context Engine for my Models 4. Replicate - Content Generation & Training Image & Video Models 5. Resend - API managed emails Also, recently - its Klipio, My own tool for Static Image ad creatives from competitor ads. Comment from Justin Verlin(@justin_verlin): Claude Code VS

... galdayan): our setup ended up being a Slack webhook per agent session, posted to a channel only that agent writes to. sounds basic but the key part was making the notification say WHY it's waiting (needs approval vs finished vs

... verification for agent-generated code is still manual and ad hoc. i run multiple coding agents in parallel on different features and the only thing that consistently catches issues is treating the diff like a code review, not a rubber stamp. i read every file changed, check for hallucinated imports, verify the logic actually matches what i asked for. one pattern that helps: keep agent tasks small and scoped. 5000 lines in one shot is where things go sideways ...

... smart" to be useful. Comment from Navanita Devi(@navanita_devi): Dedy's point is interesting, the opacity problem is real. Most people conflate "AI having access" with "AI having unchecked control." They're very different things. Read-only access vs write access, logged actions vs

... hiring a Senior Software Engineer, remote EU @Mastra is hiring a Core Engineer, remote US Keep reading The Breakpoint [2026-01-07] - Happy 2026 The Breakpoint [2025-12-17] - Best AI automation tools? The Breakpoint [2025-12-10] - Rubber duck everything Comment from fmerian(@fmerian): @ilaiszp  exactly :) Comment from Ilai Szpiezak(@ilaiszp): @fmerian   Why we're building @Pretty Prompt ! Comment from Ilai Szpiezak(@ilaiszp): Love the prompt engineering thread! My new name for it: Context Engineering ...

... Developer Experience Engineer, remote WW @Thesys is hiring a Developer Advocate, in Bengaluru, IN @Mastra is hiring a Product Engineer, remote US Keep reading The Breakpoint [2025-12-17] - Best AI automation tools? The Breakpoint [2025-12-10] - Rubber duck everything The Breakpoint [2025-12-03] - Claude workflows Comment from Arpan Garg(@arpansac): It's amazing to see the lightning speed from idea to v1 for products these days, while I'm still learning by playing around, getting ...

... Hiring? Looking for work? [Startup Roles November 2025] Hiring? Looking for work? [Startup Roles October 2025] Hope it helps! Keep it up [1]: The Breakpoint [2025-12-17] - Best AI automation tools? [2]: The Breakpoint [2025-12-10] - Rubber duck everything Comment from Alex Cloudstar(@alexcloudstar): Hey PH folks I am exploring new opportunities and would love to work with a startup that is building something meaningful. Background & skills Senior JavaScript developer with ~7 years of experience Full-stack ... ... days I mostly write or ghostwrite, I also have hands-on experience with SEO research. That means I can do keyword research, create topic clusters, map these topics to the marketing funnel (ToFu, MoFu, BoFu), and even do SEO-competitor

... JavaScript toolkit, got recently acquired by @Anthropic @Unosend is an API-first, cost-effective email platform @Pylar securely connects your entire data stack to any AI agent @8bitcn/ui turns @shadcn/ui into an 8-bit styled component library Rubber duck everything Here's the thing: Sometimes it's good to take a step back and reflect rubber

... Paris, FR @Mintlify is hiring a Design Engineer, in San Francisco, CA @Kilo Code is hiring a Forward Deployed Engineer, Remote EU @Linear is hiring a Senior / Staff Product Engineer, remote EU Keep reading The Breakpoint [2025-12-10] - Rubber duck everything The Breakpoint [2025-12-03] - Claude workflows The Breakpoint [2025-11-25] - Did Anthropic win the AI coding race ...

... tools on Product Hunt. The latest Recent dev-first products launched on the site @Claude by Anthropic released Opus 4.5, their best model for coding, agents, and computer use @nao is an AI-powered IDE for data work @Rubber Duck reviews your iOS apps before you submit them to the App Store @Haxiom is a collaborative MarkDown editor @Agenta launched an open-source LLMOps platform for building reliable AI apps Claude workflows that work We talked about Anthropic last ...

... designing, thinking, solving. Developers become architects of thought, not bricklayers of logic. Programming is becoming a feedback loop between human intention and AI interpretation. It s not about perfect prompts. It s about interactive thought refinement. Claude is your rubber duck if your duck had 200k tokens of memory. So here s the question: Would you rather debug code or debug ideas? With Claude, we re choosing the latter. And once you try it, you won t want ...

... paying users? Comment from Nika(@busmark_w_nika): @tilmaninc  It doesn't mind that I asked before :) I am happy you shared your creation! :) Comment from Nika(@busmark_w_nika): @fbajo  Me: I am using Claude in VS Code. What is the difference here? Comment from Nika(@busmark_w_nika): @maryna_klokova  okay, I abandoned it because it wasn't a thing anymore  I mean: it started being less popular on the platform :D BTW, interesting ... ... template (mostly meme or reposting pages) A Chrome extension that sorts all the reels/tiktok videos by views/likes/comments on any profile. (used for content inspiration) Never launched these two, but will do it soon. Currently building a free alternative

... various parts interconnect. So I now mostly use AI for smart autocomplete, building isolated functions/modules that can be defined and tested in isolation and are not too unique for LLM to get confused, and also as a rubber duck to bounce ideas, brainstorm. Comment from Ajin Sunny(@ajinsunny): Nothing so far for me, if you are on the 3.7 Max Sonnet. It solved pretty much most with thinking. Comment from Hyuntak Lee(@hyuntak_lee): @ambika_vaish  Yes exactly! Giving ...