All activity
Mikita Aliaksandrovichleft a comment
Definitely will try it out as the automated deployments is the most interesting part right now in the era of vibe coding! Have a great launch!
DiploiGo from zero to a live full-stack app with 3 clicks
Mikita Aliaksandrovichleft a comment
Interesting angle. A lot of people can build agents now, but packaging them into something users can actually pay for is where the real business starts. Making distribution + billing simpler is a strong move. Congrats on the launch 🚀
CrossnodeVibe code AI agents and put them behind a payment wall
Mikita Aliaksandrovichleft a comment
Right now mine is pretty lean: Claude Code for actual implementation Cursor when I want a more visual workflow ChatGPT for reasoning / tradeoff thinking Good docs inside the repo so the model has context The biggest shift for me is this: the stack matters less than whether the tool can operate inside a real workflow without creating cleanup work later.
What’s Your Vibe Coding Stack in 2025?
Mikita Aliaksandrovichleft a comment
I usually think about it like this: Free should let people feel the outcome once. Paid should let them rely on it repeatedly. For AI products especially, free can prove the workflow. Paid is where scale, frequency, and real delegation start.
How do you decide what features should be free and what should be paid?
Mikita Aliaksandrovichleft a comment
One thing I keep coming back to: the interesting shift is when this stops being a tool only for engineers. If a founder, PM, or operator can describe a scoped change in plain English and still get a reviewable result, that feels much closer to delegation than assistance. That’s the category I’m personally most interested in.
What would make you trust an AI developer on your codebase?
Mikita AliaksandrovichJoin the discussion
Mikita Aliaksandrovichleft a comment
Congrats on the launch 🚀 Really important problem to solve!
CekuraObserve and analyze your voice and chat AI agents
Mikita Aliaksandrovichleft a comment
Strong angle. What’s interesting is we’re starting to see AI move from assistive tools to packaged execution. MindPal + Lovable makes productizing expertise much easier. We’re thinking about a similar shift from another angle with Ovren — not just generating tools, but letting teams delegate real software tasks to AI developers and ship scoped work faster. Feels like the market is clearly...
MindPal + Lovable is a killer combo for productizing expertise into sellable AI tools
Tham (Sylvia) NguyenJoin the discussion
Mikita Aliaksandrovichleft a comment
I’m increasingly convinced the right move is: engineer the core, vibe-code the edges. Protect the critical path, tracking, and anything expensive to rework - but don’t build a “Series A codebase” for a product with zero users. That balance is probably where most strong MVPs live.
Building SaaS in 2026: Are you vibecoding your own product or engineering it the "old way"?
Aleksej VukomanovicJoin the discussion
Mikita Aliaksandrovichleft a comment
I think hybrid is becoming the default for serious AI products. A flat SaaS fee is great for reducing friction, but once usage maps directly to infra / model costs, pure subscription pricing starts breaking down fast. For something like Ovren, I’d much rather see: base platform fee + usage / execution credits + clear limits That keeps onboarding simple, but still aligns pricing with real cost...
Is usage-based pricing becoming the norm for AI tools?
Jake FriedbergJoin the discussion
Mikita Aliaksandrovichleft a comment
This is the kind of push founders need! You can polish forever, but real signal only starts once you launch. We’re preparing Ovren too, so this definitely hits home.
Launch tomorrow and you could get a YC interview
Aaron O'LearyJoin the discussion
Mikita Aliaksandrovichleft a comment
I trust AI agents with execution, not with irreversible decisions. Research, drafts, repetitive implementation, structured workflows? Yes. Payments, legal commitments, private health data, confidential relationships? Not without hard boundaries. Building in this space made me believe the winning products won’t be the most autonomous ones. They’ll be the ones with the best trust model.
How much do you trust AI agents?
Mikita Aliaksandrovichleft a comment
Great question. My rule is usually simple: free = enough value to create trust paid = the part that creates direct outcome, saves serious time, or removes complexity People should be able to feel the value before paying. But the real transformation / execution is where paid usually starts. For builders, I think the hardest part is not “what can I charge for?” It’s what should stay frictionless...
How do you decide what features should be free and what should be paid?
Mikita Aliaksandrovichleft a comment
Looks really promising and could save few sprints for developers!
OpenFlagsFast, self-hosted, edge-ready feature flags for modern teams
Mikita Aliaksandrovichleft a comment
Big one for me: design system awareness + real product consistency. Not just generating a nice screen, but understanding existing spacing rules, typography scale, reusable patterns, and when something looks off in a real app. For AI tools, “generate UI” is easy. “Make it feel like the same product” is still the hard part. That’s where it starts becoming actually useful for teams, not just...
What design skill should our agent learn next?
Mu LiJoin the discussion
Mikita Aliaksandrovichleft a comment
My instinct is tests or low-risk refactors first. Small bug fixes are tempting too, but they can still touch more logic than expected. Curious where others would start.
What’s the first real task you’d trust an AI developer with?
Mikita AliaksandrovichJoin the discussion
Mikita Aliaksandrovichleft a comment
Looks very cool and promising, congrats @hanh_nguyen
FractalThe fastest way to ship exceptional ChatGPT apps
Mikita Aliaksandrovichstarted a discussion
What’s the first real task you’d trust an AI developer with?
Curious where the line is for people here: If you had to delegate just one real task today, what would you trust an AI developer with first? Bug fix, refactor, tests, docs, or a small feature?
Mikita Aliaksandrovichleft a comment
Very interesting direction! I will definitely give it a try!
Claude Cowork ProjectsTasks, context, and files organized in one workspace
Mikita Aliaksandrovichleft a comment
Very strong insight. AI can generate a polished competitor in minutes, but it can’t compress the trust, distribution, and content depth needed to actually win in the market. A lot of founders will underestimate that.
I asked AI to Build a Competitor to My Own Product. It Did. Here’s What I Learned.
Imed RadhouaniJoin the discussion
Mikita Aliaksandrovichstarted a discussion
What would make you trust an AI developer on your codebase?
Curious how builders here think about this: Would you trust an AI developer to work on a real production codebase today? Not just autocomplete or code suggestions, I mean actually taking a scoped task and shipping code updates. What would make you comfortable trying that, and what would be the biggest red flag for you?