Dear makers, do you keep somewhere the AI decision logs and what do you do with them? Say I have a AI chat and I ask it - create me this and that and they I use the results for something that brings bad user experience. Do you keep a log for the decision tree or the process and why?

Apart from your own use, is there a Law somewhere in the world that requires you to do that?

That's the ultimate founder dilemma. Honest answer: I honestly don't know. I'm completely split.

Part of me says NO, keep going. A Product Hunt launch is just validation that 100 early adopters exist. It proves your landing page at least works. But a directory spike isn't a sustainable channel; it's a vanity boost. Quitting just because you need a real distribution strategy is too premature.

But another part of me says YES, stop and rethink. If you can get 100 people to look at a solution to their problem and not a single one converts or comes back, then perhaps you aren't solving a real problem. Or you are, but in a way nobody wants to pay for or even use. In a market this competitive, zero organic interest post-launch is a very scary signal that you are building something nobody wants.

So, do you stop? I wouldn't stop the problem, but I would seriously question if the product I launched is the right solution.

Hi I am Anurag yadav and today Introduce my AI doesn t fail because it s weak.

It fails because it doesn t understand humans.

Today s AI predicts.

Today s AI automates.

We recently shipped our first browser extension, and getting the same MV3 build approved on Chrome, Firefox, and Edge turned into three very different experiences. Sharing a few notes in case they save someone else some time.

Same code. Three stores. Three completely different review experiences.

Wired up a connectivity check on a health app I am building and hit a gap that felt obvious once I saw it. The OS reports "connected to wifi" the moment it associates with an access point, before DNS or the actual gateway path is confirmed. So a captive portal at an airport, a hotel router with no WAN, a home router that lost its upstream link, all read as online while every real request quietly times out.

Ended up not trusting the OS flag at all. The check now does a real DNS lookup against a known host with a short timeout before marking anything online, and only flips the app into offline mode once that actually fails. Adds a few hundred milliseconds to every check, worth it for not showing a broken screen while telling the user everything is fine.

With multi-agent work, I kept seeing two extremes: script every step, or give agents a goal and watch the original intent get lost between handoffs.

Permissions and approval gates govern actions. The harder question is what an agent is allowed to decide while doing the work.

Un proveedor que siempre dice s puede ser una mala se al

Hay algo que me ha empezado a preocupar cuando hablo con proveedores:

Qu significa realmente que tres proveedores reciban el mismo brief y los tres respondan s , podemos hacerlo ?

Well I do, and after working on one the better half of last year I have started the preparation phase. What I have done so far is collect all my conversations from last year - which were many, add them to NotebookLM and start asking questions. That was very fruitful at frist, but after a while I had the feeling I was just hitting the surface. Surely I extracted maybe 30 pages of data or more, but since I worked like a real architect during those conversation there was a lot of detail included.
Eventually I ended up asking notebook to give me the name of the files that included everything about how to build the AI and let ChatGPT read them one by one to extract the details. That made all the difference. Without this we would have built a galvanised version of the old transformer system. That is not what it is going to become.

So what are your thoughts about a private local LM? What would you prefer? an assistant, a friend or 'something else'?

Shady guy- claims to be a top hunter but is here only for scamming genuine makers. He takes money to hunt a product with some silly claim of refunding the amount if the product doesn't make it to top 10 on the day of launch, and then will ghost you, keep shifting launch deadlines at his whim, and take no accountability after launch. And of course the money would never come back.
Just wanted to beware the community before anyone considers him as a hunter.

This is also a feedback on the product hunt platform, as to how some good communities that start with good intentions, end up being gamed because of their own rules and users become victims.