Most prompt tools paraphrase your idea and hand it back. Archly interviews you a 7-step intake flow that pulls out the details you didn't know mattered, then engineers a prompt built for your specific platform: Lovable, Bolt, v0, Cursor, Claude Code, Replit, and more. Add your Profile DNA (your stack, your style, your preferences) and every prompt comes out already knowing how you build. It's not a prompt generator. It's a prompt workspace, reusable skills, saved context, prompts that ship.
hey product hunt 👋
last year i launched PromptCraft here. it hit #13, got 681 signups... and 2 paying customers. ouch.
but that taught me something. people didn't want another prompt generator. rewording your idea into a prettier paragraph doesn't fix anything. the real problem is your idea is missing details you don't even know it's missing. so the AI builder just guesses.
so i scrapped it and rebuilt from zero. that's Archly.
instead of generating, archly interviews you. it asks the questions that pull out what your idea actually needs. the edge cases, the data model, the stuff you only notice after your third failed generation. then it builds the prompt for your specific platform, because a prompt that works in Lovable won't work in Cursor or Claude Code.
there's also Profile DNA. you tell archly your stack and how you build once, and every prompt comes out already knowing you. plus reusable skills so you're not repeating the same context every time.
i built this solo because i was tired of blaming the AI builder when the real issue was what i was feeding it.
free tier is live, no card needed. i'll be here all day. roast it, tell me what's missing, ask me anything.
what's the worst thing an AI builder has ever given you from a prompt?
Report
the platform-specific angle is the part that stands out to me, most tools treat "write me a prompt" as one universal problem when the actual constraints are different per builder. curious about the intake though: if someone picks Lovable but describes something that really needs backend auth flows or a real database, does archly ever push back on the platform choice itself, or does it just take what you picked and build the best prompt it can for it regardless of fit
@galdayan Honestly? Right now it takes the platform you picked and builds the best prompt for it, it doesn't push back on the choice itself, but will do that in the future version
Report
@princeajuzie fair, and honestly the more interesting product might be the pushback version. a solo dev with a client already committed to a stack doesn't want a tool arguing with them, but a founder still deciding does. when that version ships, would it be a hard flag ("this really needs a backend, reconsider") or would it build the best-effort prompt anyway and just attach a warning alongside it? the second seems safer since killing someone's momentum over a technical concern they may already know about could just get the tool ignored.
Report
The 7-step intake caught me off guard, I figured it would just be busywork but it actually pulled out platform specifics I never would have added myself. Profile DNA is a smart touch too, saved me from retyping my stack every time.
@iokhan89728 this is exactly what i was hoping people would feel tbh. the intake was the riskiest part of the build, most people bounce off multi-step forms so hearing it pulled out stuff you wouldn't have added yourself is huge. appreciate you trying it.
Evra
the platform-specific angle is the part that stands out to me, most tools treat "write me a prompt" as one universal problem when the actual constraints are different per builder. curious about the intake though: if someone picks Lovable but describes something that really needs backend auth flows or a real database, does archly ever push back on the platform choice itself, or does it just take what you picked and build the best prompt it can for it regardless of fit
Evra
@galdayan Honestly? Right now it takes the platform you picked and builds the best prompt for it, it doesn't push back on the choice itself, but will do that in the future version
@princeajuzie fair, and honestly the more interesting product might be the pushback version. a solo dev with a client already committed to a stack doesn't want a tool arguing with them, but a founder still deciding does. when that version ships, would it be a hard flag ("this really needs a backend, reconsider") or would it build the best-effort prompt anyway and just attach a warning alongside it? the second seems safer since killing someone's momentum over a technical concern they may already know about could just get the tool ignored.
The 7-step intake caught me off guard, I figured it would just be busywork but it actually pulled out platform specifics I never would have added myself. Profile DNA is a smart touch too, saved me from retyping my stack every time.
Evra
@iokhan89728 this is exactly what i was hoping people would feel tbh. the intake was the riskiest part of the build, most people bounce off multi-step forms so hearing it pulled out stuff you wouldn't have added yourself is huge. appreciate you trying it.