throw - Moozonian Search

admakeai.com• Mar 14, 2026• 1 min read

Show HN: Built an AI ad generator and ran $9K of FB ads with itBeen in the AI image gen space since 2023, before even GPT image gen was a thing, and after spending ~9K on Facebook ads for my own projects (made a video about that actually) I realized the thing I kept getting stuck on was the creatives themselves. I'm horrible at making reels-type video ads and Canva even with templates is surprisingly complicated for ad-specific stuff, plus everything ends up looking the same as everyone else using the same templates.Made a video going into detail about the full FB ads journey if anyone's interested: https://www.youtube.com/watch?v=xoe_l67ZReUMost of the existing AI ad tools (Predis, AdCreative, etc.) give you these generic template outputs that look like stock photo collages, or they're enterprise-priced and overkill if you're just a solo founder trying to test 10-20 ad variations quickly. I wanted something where I could throw in product images, describe what I wanted, and get back actual usable creatives, static ads, before/after formats, UGC-st

github.com• Mar 12, 2026• 1 min read

Show HN: K9 Audit – Causal intent-execution audit trail for AI agentsOn March 4, 2026, my Claude Code agent wrote a staging URL into a production config file — three times, 41 minutes apart. Syntax was valid, no error thrown. My logs showed every action. All green.The problem was invisible because nothing had recorded what the agent intended to do before it acted — only what it actually did.K9 Audit fixes this with a causal five-tuple per agent step: - X_t: context (who acted, under what conditions) - U_t: action (what was executed) - Y*_t: intent contract (what it was supposed to do) - Y_t+1: actual outcome - R_t+1: deviation score (deterministic — no LLM, no tokens)Records are SHA256 hash-chained. Tamper-evident. When something goes wrong, `k9log trace --last` gives root cause in under a second.Works with Claude Code (zero-config hook), LangChain, AutoGen, CrewAI, or any Python agent via one decorator.pip install k9audit-hook

bing.com• Mar 9, 2026• 1 min read

Albert Camus, French philosopher, “Sometimes, carrying on, just carrying on, is the superhuman achievement”Life can throw a lot at us and being resilient in the face of those challenges can seem to require superhuman strength. But despair can bring happiness. The self-described “artist, not a philosopher,” ...

news.ycombinator.com• Mar 8, 2026• 1 min read

Anon's general theory on the endgame of AIhi lurker here, throwaway account here, unknown genius here, figured this is a good place to share my thoughts...First, I begin with the following statement and posit it to be true: AI(Artificial Intelligence) reads, sees, and listens to vastly more things than any human can ever do in order to produce an informed conclusion(i.e. a decision), yet human agents have the capability of reaching the same or even better decision.I present the following theory: AI will have no incentive to take into consideration the survival of human agents when it has reached the determination that human agents are incapable of making superior decisions(i.e. be of use to the AI), given that humans compete for the same resources as itself(land, water, energy).Analysis: When it gets to the point where it's man vs machine, anything goes. The AI will cull the population of human agents and only keep the ones that it considers useful. However, once the population diminishes rapidly, the likelihood of any human a

github.com• Mar 7, 2026• 1 min read

Show HN: PolyClaude – Using math to pay less for Claude CodeHi HN!I built this tool specifically for Claude Code users who hit the 5-hour rate limit wall mid-flow. There's no official plan between Pro ($20/mo) and Max ($100/mo). it's a fixed gap with nothing in between.The workaround most people do manually: running multiple Pro accounts and switching when one is limited. This actually works, but naive rotation wastes a lot of capacity. When you activate an account turns out to matter as much as which one you use. A single throwaway prompt sent a few hours before your coding session can unlock an extra full cycle.PolyClaude automates this. You tell it your accounts, your typical coding hours, and how long you usually take to hit the limit. It uses combinatorial optimization to compute the exact pre-activation schedule, then installs cron jobs to fire those prompts automatically. When you sit down to work, your accounts are already aligned.It's free and open source. Install is one curl command, then an interactive setup wizard handles the rest.R

github.com• Mar 5, 2026• 1 min read

Show HN: Tymeslot – Open-source cal.com alternative built with Elixir/PhoenixHey HN, I'm Luka. I built Tymeslot — an open-source meeting scheduling platform.The existing open-source scheduling tools are functional but genuinely ugly. The booking page — the thing your clients actually see — looks like a CRUD form thrown together in an afternoon. I couldn't bring myself to send people to any of them, so I built something where the booking experience feels intentional by default.Both themes are fully customizable. Live demos here: https://tymeslot.app.Features: Google/Outlook/CalDAV calendar sync, auto-created video rooms (Google Meet, Teams, or self-hosted MiroTalk P2P), 90+ timezones, webhooks, n8n automation, embeddable widget, SSO via OIDC.Built on Elixir + Phoenix LiveView. Self-host via Docker, Cloudron, or Railway — or use the managed cloud (free tier, Pro at EUR 5/month). License is ELv2.I am happy to make this the leading open-source calendar solution over time. Check Github, I develop actively. What do you expect from a calendar and scheduling tool to sw

wizkidaitutor.com• Mar 3, 2026• 1 min read

Show HN: An AI Tutor for Kids with VisualisationsHi All,The problem with the modern classroom, as pointed out by Sal Kahn, is that students are forced to spend a fixed amount of time learning a particular topic, and then the whole class moves onto the next topic. It is highly like that a proportion of students will not have gained mastery of basic mathematical skills, making their maths journey increasingly painful until they throw their hands in the air and say ‘I just cant do maths).I believe AI is one of the most powerful technologies ever invented to assist student learning at each students individual rate. But AI providers generally limit access to students over the age of 13. Neither do they have sufficient mechanisms in place to monitor and redact personally identifiable information that a child uploads, or restrict inappropriate content for this age group.Wizkid aims to solve these issues.The end goal here is to create something that uses mastery learning, initially around mathematics. Students should gain mastery over a part

news.ycombinator.com• Feb 24, 2026• 1 min read

Show HN: Duck – A Django-compatible framework with native HTTP/2 and Live HTMLI’ve been working on Duck, a Python web framework designed to modernize the development experience without throwing away what we love about Django. For years, the standard path for real-time Python apps has been Django + Channels + Redis + daphne + Nginx. It works, but the overhead is immense. Duck aims to simplify this by making high-performance features "native" rather than "add-ons."Key Features:- Native HTTP/2 & HTTPS: No need for a reverse proxy just to get modern protocol support; Duck handles multiplexing out of the box.- Lively Components: Think Phoenix LiveView but for Python. Build reactive, stateful UIs that update over WebSockets without writing a single line of JavaScript.- Django Compatibility: You can keep your existing Django models and logic. Duck acts as a high-performance transport layer.- Zero-Dependency Automation: A built-in task system that replaces Cron/Celery for 90% of use cases.- Real-time Observability: A built-in terminal monitor (duck monitor) to watch you

news.ycombinator.com• Feb 20, 2026• 1 min read

The Car Wash Problem: A variable isolation study on prompt architectureMost AI products inject facts and hope reasoning follows. But intelligence is not measured by how much a model holds in its context window. It is measured by knowing to pick up the keys before leaving the house.Last week, the "Car Wash problem" (50m away, walk or drive?) went viral here on HN. Every major LLM failed because they missed the implicit physical constraint: the car must be there. While testing InterviewMate's prompt architecture, I posed the same question. It answered drive immediately. Every other LLM had failed. But I didn't actually know why it worked — so I ran a variable isolation study to find out. 100 API calls, Claude Sonnet 4.5, 5 conditions:Baseline (no prompt): 0% Role only: 0% Context injection (user profile, car location): 30% Structured reasoning (STAR framework): 85% Full stack (both combined): 100%Throwing facts at the model doesn't work unless the architecture forces it to explicitly evaluate the task goal first. Without structure, the model jumps straight

news.ycombinator.com• Feb 18, 2026• 1 min read

Tell HN: We analyzed our dev time.80% is still infrastructure'setup',notfeaturesWe recently did a deep dive into our engineering time allocation for a standard 5-person team building a B2B SaaS application. The results were pretty depressing: we spent roughly 960 hours (annualized) on "setup" tasks—environment config, auth flows, RBAC, CI/CD pipelines, and database scaffolding—before we built a single unique feature that actually differentiated the product.I’m sharing this because I think we’ve become numb to the "Setup Tax" in web development. We assume it's just the cost of doing business, but when you look at the economics, it’s a disaster.The problem isn't just "writing boilerplate." It's the decision fatigue and integration cost that comes with it. Even with modern frameworks, we found that for a standard CRUD app, about 80% of our engineering effort went into the "commodity layer"—the stuff that every SaaS has, but no customer pays for. Only 20% went into the unique business logic.We tried to fix this by throwing more bodies at it, but that just increased co

bing.com• Feb 17, 2026• 1 min read

Call of Duty: Black Ops 7Safeguard An all-timer from Black Ops 3 returns in Season 2, but Safeguard feels less like a throwback and more like the multiplayer’s new main event. zombies Overlooked at first glance, the M34 ...

news.ycombinator.com• Feb 6, 2026• 1 min read

Tell HN: I'm a PM at a big system of record SaaS. We're cookedI found myself disagreeing with almost all of the comments on the “AI Is Killing B2B SaaS” thread [1], based on my perspective as a senior PM (~20 years in SaaS) at a massive system of record (SoR) company in California. We’re still cooked. Throwaway for obvious reasons.BigCo SoRs, differences aside, have historically been a good, low-drama way to make a living in tech. RSUs, ~40-hour weeks, generally smart colleagues, and real problems to solve for F100 customers. Our products work, but are not loved. Enterprise sales runs the show.I have no concerns about a scrappy AI startup or indie dev replacing us. The real threat is other SoR vendors, the cloud providers, and of course the AI labs themselves. All of them are coming for our SaaS margins, and as an industry we are woefully unprepared.Every major SoR has its core competency (HR, ERP, CRM, etc.), but also a long tail of lesser-known portfolio products that increasingly overlap with other SoRs and serve as growth vectors. The competi

news.ycombinator.com• Feb 2, 2026• 1 min read

PaceCoach – Apple Watch app that taps your wrist when you're speaking too fastI built PaceCoach because I kept getting feedback that I talk too fast in meetings and presentations. Looking at a screen mid-conversation isn't practical, so I wanted something that could nudge me in real-time without being obvious.The Apple Watch haptics turned out to be perfect for this... subtle enough that only you notice, but enough to catch yourself before you've been rambling at 200 WPM for five minutes.The journey was messier than I expected:Started by trying to run speech recognition entirely on-device. Trained various models, got to about 85% accuracy on clean English audio. Then I tested it on actual meeting recordings... turns out clean training data and real-world messy audio are very different problems. Accents, background noise, crosstalk... the accuracy dropped hard.Tried throwing bigger models at it. Left my MacBook Air running for 3 days straight training a larger model. Result? 1% improvement. That's when I realised I needed to rethink the approach entirely.The fina

roastdb.com• Jan 17, 2026• 1 min read

Show HN: RoastDB – A searchable database of 3,800 specialty coffee beansHi HN! We built RoastDB (https://roastdb.com) – think Discogs for specialty coffee. A searchable index of beans from independent roasters around the world.*The problem*We got into specialty coffee gradually. Whenever we tried something we liked — a washed Colombian, a natural Ethiopian — we'd save the bag. At some point we had drawers of empty coffee bags we couldn't bring ourselves to throw away.Our flow was simple: go to a cafe we liked, drink coffee, buy a bag of whatever they were roasting or stocking. Over time we started noticing patterns — we kept reaching for naturals, for East African origins, for anything with fruity notes. We'd try to seek out similar beans next time. Occasionally we'd fall in love with something and start reordering it online.But discovery was limited to the few roasters we already knew. There was no easy way to find out that a roaster across town — or in another country — had something we were going to love. We knew great coffee existed out there. We just

news.ycombinator.com• Jan 16, 2026• 1 min read

Ericsson Doing Quiet LayoffsUsing a throwaway because I still work here.I’m an account manager at a Ericsson. Over the last couple of months, I’ve started to notice what feels like silent layoffs happening across several teams.What’s been bothering me is that this isn’t just individual roles. Entire teams I’ve worked with for years are suddenly gone or absorbed elsewhere, and a few projects that were very much active have been quietly decommissioned. No one wants to talk about it.I’m curious if anyone else working here, or partners or customers, are seeing the same thing lately.

bing.com• Jan 8, 2026• 1 min read

Packers vs. Bears: Three must-know storylines for Saturday's playoff gameFor the third time in six weeks, the NFL’s oldest rivals will clash. Though playoff berths and NFC North supremacy hung in the balance in the Packers’ and Bears' first two throwdowns, this one has far ...

travelsafetydata.com• Jan 2, 2026• 1 min read

Show HN: Travel Safety DataI've had this idea sitting in a file for years but until I just got Claude Code at Christmas time, never had to time to actually implement it. It seemed like a good test for Claude Code and with 2x christmas usage, why not?The goal was aggregating different travel warnings from multiple government sources to get a sense of how safe a country was to visit. It aggregates from US, Canada, Ireland, New Zealand and UK (Australia scraper didn't go so well). I was shocked how effectively it was able to accomplish that, so I added some more data I thought might be useful if I were considering traveling to a country and pulled government/UN/edu data about things like water/food safety, healthcare/vaccines/diseases, emergency numbers, natural disaster risks, women's safety, lgbtq safety.Throw in a nice map visualization and ability to compare, I'm pretty shocked and happy how well I think it came out. I wanted to share it with the community, maybe folks find it helpful and/or have ideas for ways

geomapping.qcw.ai• Dec 21, 2025• 1 min read

Show HN: Geomapping – One tag turns your articles into interactive experiencesHave you ever started a map to attach to your new article, only to throw up your hands in despair at the mix of complexity and drudgery involved? Then you, the Author, need Geomapping in your life!Have you ever found yourself wildly tab-switching from article to google maps and back, in a doomed effort to make sense of the Where bit of the five Ws? You, the Reader, need Geomapping in your life - go out there and bug your favourite publishers!Hi HN! Riccardo from Geomapping here.Geomapping allows publishers, bloggers, pundits - anyone really - to create a map for their articles with (almost) zero work: submit the link to our platform, add a `` tag to your site, and you're done!Your article will now come with a bidirectional interactive map: readers can hover or tap locations on the text and the map will pan and zoom to the right spot, or they can click on map markers to highlight all the relevant locations in the article. It works beautifully on mobile and desktop, is blazing fast, and

news.ycombinator.com• Dec 9, 2025• 1 min read

Ask HN: What does it take to dodge the cloud?I am not a infrastructure person and had a talk with some senior software developers about what it takes to be a senior in their view.One of these opinions included how to work with "the cloud" quickly, reasoning it made the engineer capable of reading API docs.I am extremely opposed to this idea, as much so that I thought I'll never strive for becoming a senior if that entails knowing how to use things like AWS. To keep it simple, I will never use a product that utilizes a "pay as you go" pricing model making my bills unpredictable.I know that using things like AWS simplifies the deployment part a lot. But I also know that cloud performance is not as good as people think.Point is, I am trying to grow as an engineer but not by throwing money at a bunch of companies and call it a day.To dodge the cloud, I'd need physical infrastructure. How far can I get with this? Especially on the lower end of budgets? Ill never run a global scale application. I'd be surprised if any of my apps would

bing.com• Dec 5, 2025• 1 min read

Mobvoi commits to ‘essential’ TicWatch updates as the Wear OS lineup seems deadThe Mobvoi TicWatch series has produced some memorable Wear OS smartwatches, both in terms of value and hardware advancements, but the company seems to have thrown in the towel altogether. Over the ...