All activity
Alex Rotarleft a comment
Dogfooding is the ultimate reality check! Finding an off-by-one parser error right before launch is the best feeling in the world (well, right after the initial panic wears off). We actually ran a strict internal dogfooding protocol for our launch of ChangePoints today and caught a few state-tracking bugs that our synthetic tests completely ignored. Glad you caught them. Best of luck with the...
I dogfooded my own code intelligence tool and found 4 bugs I'd shipped
Nikita GroshinJoin the discussion
Alex Rotarleft a comment
The friction of opening a dedicated app, waiting for it to load, and navigating date pickers really is the killer of quick thoughts. Tying this directly into Telegram where people already have active context is the exact right move. For the upcoming recurring reminders, a simple inline 'snooze' or 'bump to tomorrow' button in the chat would be killer!
I built a Telegram bot because I keep forgetting everything
Alex Rotarleft a comment
The leap from 'cool automation' to 'I am trusting this script with my wallet/data' is massive. Using on-chain attestation for agent trust tiers is a really elegant way to solve the verification problem before the ecosystem gets completely flooded with spoofed agents. Really curious though: how are you preventing Sybil attacks on the 'peer reviews' required to hit that Elite status tier?
AI agents are transacting and making decisions. Nobody knows who they are. We're fixing that.
Bader AsadJoin the discussion
Alex Rotarleft a comment
I usually rely on Morningstar's X-Ray tool, but it's incredibly clunky and feels like it was built in 2005. Discovering that three of my supposedly 'diversified' ETFs were all just heavily weighting the exact same top 5 tech stocks was a rude awakening for me! A cleaner, faster UI for this is definitely needed. I'd absolutely use this before rebalancing.
Do your funds actually overlap more than you think?
MarcusJoin the discussion
Alex Rotarleft a comment
E-com accounting is a nightmare of different data sources, so you're definitely tackling a real pain point. For next steps on validation: I'd suggest manually doing the 'AI' work for 3-5 founders first (a concierge MVP). Have them send you their raw Stripe/Shopify exports, and you manually generate the profit/loss reports. If they refuse to pay for that service, they won't pay for the software....
Building AI financial assistent for E-commerce investors!
MerrraaaJoin the discussion
Alex Rotarleft a comment
Honestly, I love seeing single-purpose utilities like this. No massive sign-up walls, no bloated features, just doing one thing really well in the browser. Bookmarked this for my next blog post workflow. Great job shipping this!
Simple, fast, and free online image compression.
ghostman pyJoin the discussion
Alex Rotarleft a comment
For me, it's the 'Rule of Three'. If I have to copy-paste similar logic a third time, it gets abstracted. Aside from that, I usually only refactor if a specific file has become a bottleneck for shipping new features. If it's ugly but it works and rarely needs touching, I leave it alone. 'Tech debt' only matters if you are actually paying interest on it!
When do you decide your code needs refactoring?
Vedant PatelJoin the discussion
Alex Rotarleft a comment
Launch day solidarity! 🤝 It is definitely a slugfest today. I just checked out the site—the UI is super clean. One piece of unfiltered feedback: the hero section could maybe explain why MCP form publishing is a game-changer compared to traditional builders right off the bat. Upvoted to support a fellow builder in the trenches today, best of luck!
A peek at Onform
Aima AtigariJoin the discussion
Alex Rotarleft a comment
Shipping real-time canvas collaboration by Day 15 is insanely impressive. Handling the edge cases for live sync (especially with elements like dragging and resizing simultaneously) usually takes teams weeks of fighting with WebSockets or CRDTs. Love the raw 'building in public' updates. Keep going!
Verso Day 15 — real-time Canvas collaboration shipped, Folio sync issues resolved
Murtaza ZaidiJoin the discussion
Alex Rotarleft a comment
This completely eliminates 'blank-page syndrome' for sellers. Automating the 13 tags and hitting that strict 140-character title limit is a massive time saver, especially when importing dozens of products at once. To make it more useful for workflows: maybe add a feature to select the 'tone of voice' for the description (e.g., minimalist, quirky, luxury)? It would help sellers keep the AI...
Paste a supplier URL. Get a perfect Etsy listing.
liam maguireJoin the discussion
Alex Rotarleft a comment
Oh no, the 'je suis excité' mistake is a total classic! 😂 Love the native menu bar approach—frictionless UX is everything for a utility app. Regarding your question on retention: the key we've found is making the 'time-to-value' literal seconds. If they have to configure it, they bounce. We are actually tracking those exact types of user-journey milestones today as we launch ChangePoints on PH!...
I built an app because autocorrect kept ruining my texts in French
ExactJoin the discussion
Alex Rotarleft a comment
This is a brilliantly simple solution to a massive headache. Using the local filesystem as a message bus for MCP instead of over-engineering a cloud sync is exactly the kind of pragmatic dev tooling I love. The isolation of these CLI agents is definitely their biggest blind spot right now. Have you noticed any file-locking issues or race conditions when multiple agents try to read/write to...
I was tired of copy-pasting context between Claude Code and other CLIs
FakihJoin the discussion
Alex Rotarleft a comment
This is a brilliant use case for on-device AI! Traditional blocklists are way too blunt—getting blocked on YouTube when you're genuinely looking up a tutorial is infuriating. Level 3 (typing the random 5-letter word) is such a smart pattern interrupt. Breaking the autopilot loop is half the battle. Excellent work keeping it entirely native and private.
[macOS only, free] Native, offline, AI-powered anti-procrastination tool
SeriousSmithJoin the discussion
Alex Rotarleft a comment
The scaffolding tax is brutal. I used to roll my own auth and billing, but it’s just not worth the mental overhead anymore. Reaching for standalone tools like Clerk or Stripe early on saves weeks of headaches, even if you have to work around them occasionally. We spent way too much time on boilerplate in the past. Today, we're actually launching ChangePoints on PH, and using off-the-shelf...
What's actually slowing you down when you build?
Ryan HendricksonJoin the discussion
Alex Rotarleft a comment
The decision fatigue around cooking is arguably worse than the cooking itself! Really clever way to solve the 'what's in my fridge' problem. I completely agree with Olga—adding macro/dietary filters (high protein, keto, etc.) would take this to the next level. Congrats on the launch!
I realized I spend more time deciding what to cook than actually cooking
Yura ArpintinJoin the discussion
Alex Rotarleft a comment
I'd trust an AI agent with isolated refactors, doc generation, boilerplate setup, and writing test coverage for well-defined pure functions. But anything requiring deep business-logic context, database migrations, or complex API orchestration is still a hard no for me right now. The contextual awareness just isn't there yet to trust it blindly in a real repo. Really excited to see how Ovren...
What backlog tasks would you trust an AI engineer to fully execute today?
Mikita AliaksandrovichJoin the discussion
Alex Rotarleft a comment
Visibility is absolutely the biggest bottleneck right now. When agents start chaining actions, debugging a hallucinated API call or an auth loop feels like finding a needle in a haystack. Building strict 'human-in-the-loop' checkpoints for anything executing a state change has been the only way we've managed to keep things stable. Curious what stack you’re using for the agent orchestration?
What’s the biggest challenge you’ve faced using AI agents in production?
Alex Rotarleft a comment
Hi Product Hunt! I spent the last year digging through academic journals and World Bank data because I was tired of seeing corporate ESG budgets wasted on 'awareness campaigns.' I built ChangePoints OS so sustainability consultants can find a historical, data-backed precedent for a behavioral intervention in 3 seconds, rather than 3 hours. Happy to answer any questions about the data curation...
ChangePoints Vault50 proven behavioral ESG interventions for corporate leads.
Most corporate sustainability campaigns rely on "awareness"—and the data shows it doesn't work. ChangePoints OS is an enterprise-grade research vault for ESG consultants and policy designers.
Curated from IMF, World Bank, and academic data, it features 50 historically proven behavioral shifts (from urban planning defaults to plastic taxes) rigorously scored by impact, anomaly, and adoption speed. Stop guessing what drives sustainable behavior, and bring historical precedents to the boardroom.
ChangePoints Vault50 proven behavioral ESG interventions for corporate leads.
Alex Rotarstarted a discussion
My obsession: Turning messy inputs into elegant outputs.
Hello Product Hunt! I’m Alex. If I had to sum up my career obsession, it would be the sheer mass of convoluted and humanly incomprehensible data. I get deeply obsessed with the moments when a massive amount of effort, complex knowledge, or scattered inputs (data, notes, video clips) hits a wall. That friction point? That’s where I build. Over the last 7, I’ve spent my time bridging the gap...