If you’ve got something working but need another opinion, drop it here.
Share the link, what it does, who its for and the main thing you want feedback on. Landing page, onboarding, pricing, design, actual product - anything is fine.
Dont just dump a link though. Give enough context for people to actually help.
If you post yours, try review someone else’s aswell. Honest feedback but dont be a dick lol.
I’ve noticed I’m happy letting AI touch basically everything now - code, debugging, research, design.
But there are still certain decisions where I’d rather do the thinking myself than just accept whatever it gives me.
Not what you’re planning to build or what’s nearly ready. What did you actually get in front of people this week?
Could be a full product, one feature, a landing page or even something that completely failed lol.
Drop what you shipped, what stack you used and the main thing you learnt from it. Links are fine but actually give some context instead of just dumping them.
Ill go through everything and give proper feedback aswell.
I run a small company and I kept running into the same thing. Someone would ask how to do something in one of our systems, I'd walk them through it on a call, and two weeks later someone else would ask the same question. Every time I thought about writing it down properly, I'd do the math on screenshotting every step and typing it all out, and put it off again.
I looked at Scribe first, which is the obvious tool in this space and does the core job well. The thing I kept wanting was to just talk while I recorded. When I walk someone through a process out loud, I explain the why, not only the clicks, and that explanation is the part that actually helps the next person. I couldn't find a way to capture that, so I ended up typing the descriptions in afterwards anyway, which was the part I was trying to avoid.
So I built ScribeIt. You hit record, do the task the way you normally would, and it captures each step with a screenshot as you go. If you talk while you're recording, it turns what you said into the description for each step. When you stop, there's a step-by-step guide ready to edit, export as a PDF, or share with a link.
It works on any website that Chrome allows, which mattered to me because half our processes live in tools nobody has ever heard of.
I'm not a developer. I've spent the last few months figuring this out piece by piece and there's plenty I'd do differently. It's free to use and there's a paid plan if you need more room.
Happy to answer anything about how it works or how I built it. And if you try it and it does something stupid, I'd rather hear it here than find out from a one star review.
Here's the link: https://chromewebstore.google.com/detail/ldpciklepkaiacchoamfkfclkhngpmkj?utm_source=item-share-cb
Alright, I want to see what everyone here is actually building.
If you have a working AI product, even if it is still rough, make a standalone post in the sub with the Roast My MVP or Build Log flair.
Include what it does, who it is for, the link and the one thing you most want honest feedback on.
I want to test as many as I realistically can and give useful feedback. No need for a polished launch or a perfect landing page. I would rather see something real and unfinished than another vague coming soon page.
What have you got?
I've been experimenting with a different way of using AI.
Instead of opening a chatbot, asking it something, and leaving, the idea is to put AI agents directly inside group chats with actual people.
So you can create a group with friends and add different agents depending on what you want the group to do.
Some examples we’ve been testing:
The part I find most interesting is when the group stops treating the AI like a tool and starts treating it more like another participant in the chat.
You can have humans arguing, agents disagreeing with each other, people asking them to pull in context, and then the conversation just keeps going.
I’m curious what people here would actually use something like this for.
If you could add any AI agent to a group chat with your friends, what would you make?
We’re testing this with Sircle right now if anyone wants to play with it: https://sirclesocial.com
Would genuinely love some weird use-case ideas because I feel like there are probably much more interesting ones than the obvious productivity stuff.
Doesn’t have to be the whole startup either - could literally just be one feature that finally works 😂
Feel like we spend so much time talking about what’s broken. Show us something that’s actually going right for once lol
Open-source, local-first platform for building MCP-powered AI workflows. Model management, MCP servers, a visual flow builder, chat, and headless automation in one app.
try.flujo.com.co, check it on github and tell me how useless it is.
New week so what are you actually working on rn?
Drop what you’re building, the stack you’re using and the one thing you want finished by Friday.
Doesnt need to be a massive launch. A feature, landing page, automation or even fixing something thats been broken for ages all counts.
Come back at the end of the week and say if you actually shipped it lol.
Building Olympus — an open-source cryptographic document provenance ledger aimed at press freedom orgs (journalists can prove a document hasn't been tampered with, without it ever leaving their machine — only a proof gets published). Rust + Tauri 2, Sparse Merkle Tree ledger, ZK redaction proofs.
Solo, self-taught, working a trade job on the side, heavily AI-assisted (Claude Code, Codex, etc.) for implementation while I handle architecture and threat modeling.
Looking to trade eyes with other founders here — drop your repo below and I'll go through it, and I'll link mine back. Especially useful if you've got a security/architecture background, since I don't have a formal CS one and could use outside eyes on the crypto and threat model.
If I still do not understand why I need your product after the first minute, I probably never will.
The technology can be complicated. The value should be obvious.
How quickly could a complete stranger understand yours?
Every time I think something is basically done, I realise I could add another feature in like 20 minutes lmao
I fell into this trap because automating everything felt productive.
Then I realised I was spending hours automating tasks that I did not even do regularly.
Automate the shit you repeatedly waste serious time on.
Everything else might just be productive procrastination.
I've been building a Bible study app called Sojourn. It's a free passion project. No signup, no subscription, no ads.
The whole idea started from something that's been bugging me. We're all spending more and more time talking to AI. So I kept asking myself: what if AI pointed people somewhere better than itself?
That's all Sojourn really does. You ask what you're actually wondering about, and instead of giving you a chatbot monologue, it walks you straight into the actual text. Every verse comes from the real Bible, never made up or paraphrased by a model. Tap a verse and you're reading the whole chapter.
The goal isn't to keep you talking to AI. It's the opposite. Less time with the bot, more time actually sitting in scripture.
Learn more about the app: https://trysojourn.app/
Download on App Store: https://apps.apple.com/us/app/trysojourn/id6792011966
People call almost every new AI product a wrapper now, but that feels way too lazy.
If a product takes a model and adds a genuinely useful workflow, better UX, proprietary data, integrations, reliability and solves a specific problem, is it still just a wrapper?
Most software is built on top of somebody else's infrastructure anyway.
So where is the actual line for you? What makes an AI product feel defensible instead of something anyone could recreate in a weekend?
I keep seeing the same loop.
Someone builds a product, barely anyone uses it, and instead of talking to users or figuring out distribution they add another feature.
It makes sense because coding feels productive and marketing feels awkward. AI makes it even easier to stay in that loop because you can build the next feature in a few hours and convince yourself you're making progress.
At what point do you stop building and admit the actual problem is getting people to care?
Have you ever caught yourself coding around a distribution problem?
Everyone talks about the tools and models they use, but I feel like the actual skill behind them matters way more.
What do you think you're genuinely best at when building with AI?
Could be coming up with ideas, prompting, coding, design, automations, knowing what to cut, actually shipping, getting users or something completely different.
And what's the part you're still pretty shit at lol?
I feel like AI has made it way easier to get attached to an idea before you even know if it’s actually good.
Like you can have the site looking clean, the product mostly working, branding done, a bunch of features added… and because it all looks “real” you kinda start convincing yourself you’re onto something.
Not what you’re planning to build or what’s nearly ready. What did you actually get in front of people this week?
Could be a full product, one feature, a landing page or even something that completely failed lol.
Drop what you shipped, what stack you used and the main thing you learnt from it. Links are fine but actually give some context instead of just dumping them.
Ill go through everything and give proper feedback aswell.
I swear half the time I ship something before I even fully understand how it works 😂 Anyone else?