Self-Promotion
p/self-promotionShow off what you're working on
trending

2mo ago

Built a tiny library after our AI confidently lied about a file that didn't exist

A few weeks back our internal AI CTO assistant told me it had

fixed a bug in a specific file fake path, fake line number, said

with total confidence. That file never existed.

Turned out prompting alone doesn't reliably catch this, and having

29d ago

Shipped an app with Cursor / Lovable / v0? Get its A–F security grade — free

Quick update for anyone who followed us at launch

AI coding tools are incredible and they quietly ship security holes: a

Supabase table with RLS off, an API key sitting in the frontend bundle, a

"god-mode" endpoint with no auth, a chatbot that leaks its system prompt if

2mo ago

Early access for Trippy Visualizer launching Thursday

Hi folks, I m hoping to launch Illucid on Friday for GPT day (pending Apple s approval now)
It s a trippy visualizer (iPhone app) based on some of the work that lead to Google s Deep Dream, running in real-time and reacting to audio.
You can access it early by joining the public beta: https://testflight.apple.com/joi...
I hope you ll check it out, provide some feedback, and enjoy!

28d ago

Looking for businesses and designers who want to try my tool and give me feedback

I ve created a creative tool that resizes any banner to all the sizes needed to run online campaigns.

I m looking for businesses and designers who want to try it out and give me feedback. At the moment, there isn t a free trial available, but I ll select a few people from here who are interested and offer them free access.

I made a short video showing how easy is to resize a banner

29d ago

I built an AI planner you can call — TimeLift is live today 🚀

After six months of solo development, TimeLift is live on Product Hunt today!

TimeLift is your planner, minus the typing.

Call Tami, your AI cat assistant, and simply say:

Reply to emails, meeting at 3, gym later.

VersionMe isn't another habit tracker. Here's what makes it different.

Most habit apps do one thing track what you did.

Did you exercise? Check.

Did you meditate? Check.

Did you drink water? Check.

But tracking never told you why you kept failing. And it definitely never told you what to do next.

VersionMe works differently.

Instead of a checklist, you get a Mission. Instead of a streak, you have a Level. Instead of a template, you get a plan built around where you actually are.

The AI coach looks at your current level, your recent performance, and gives you one clear next step not twenty habits to maintain.

Not what a fitness influencer does at 5AM. What you should do. Today. At your level.

VersionMe is launching very soon. Built by one person. Solved one real problem.

Would you try an app that tells you what to do next instead of just tracking what you did?

30d ago

What wellness data do you wish lived in one place?

NutriFit started from a simple frustration: personal wellness data is everywhere, but almost never connected.

Nutrition may live in one app, body measurements in another, lab results in PDFs, workouts in a fitness tracker, supplements in notes, and conversations with a nutritionist or trainer in chats. Even when all of this data exists, it is difficult to see the full picture or understand how different parts of your lifestyle change over time.

How VersionMe's AI coach actually builds your habit plan

A lot of AI apps say "personalized." Most of them just change your reminder time.

VersionMe does something different here's exactly how it works.

Step 1: You set your goal, not your routine You tell VersionMe what you want to become. Not "I want to exercise 5 days a week." But "I want to have more energy and feel stronger."

The routine comes later. First, the direction.

4mo ago

AI demos always work. Production never does.

I've sat through a lot of enterprise AI demos. The agent always understands the question, pulls clean data, returns a polished answer in under five seconds. Impressive every time.

Then we go live.

Same agent, real environment: question full of internal acronyms and a typo, four systems to query with two of them returning stale data, three clarification loops, and eventually a handoff to a human for the edge case nobody put in the demo script. Every sing time.

The model isn't the problem. The mess around it is.

28d ago

We shipped an AI feature that refuses to run on phones. On purpose.

We build a PDF toolkit that runs entirely client side. There is no upload endpoint on the domain, so documents are never transmitted. That constraint is the whole product.

Two months ago we added on-device AI: summarize a PDF, ask it questions, translate it. The setup:

- A 1.5B parameter model running in the browser through WebGPU

- Executes on the user's own GPU, inside a sandboxed worker

First
Previous
•••
808182
•••
Next