working theory for a new vibe code workflow: 1. create a proof of concept in v0 2. move to github and clone in a local repo 3. use as ref for agent planning 4. create a backend directory to test API plumbing 5. create a frontend directory to prototype... iteratively this way you can repeatedly rip apart the frontend for ideation without worrying about breaking changes to the backend. i really like asking for an open ended "job to be done" (i.e. place a bet with a single tap) with 5-10 variations to feel. it's the modern day wireframing. it's still UI slop but i can get ideas in my head out faster and reveal design shortcomings before polishing or worrying about making components functional. and because the v0 prototype gets you to functional ai features and backend integration fastest, you have both a technical and conceptual reference to use in your agentic planning so your v1 is more predictable with minimal effort.show more

it’s so rare to like a fully stylized design from any of the models i’ve religiously been saying “make basic html, no styling, no formatting, barebones” for every inital ui/ux i make now but most of stuff i make is for scripting/internal tooling so maybe different

im still using figma for that

Similar Workflow 1. create a proof of concept in v0 2. move to github and clone in a local repo 3. Use Cursor to expand and work on it + use that to check code generated to clean up code 4. Publish using Vercel

separating g frontend from backend even if its a simple folder really helps alot

This is exactly I start the projects

the real step 6: stare at 10 variations and pick the worst one somehow

nope. ask for 10 more.

That's how i started and got my bf to vibecode! Really like this workflow

I would just use @cursor_ai or @ManusAI (if you’re less technical) to do this

Very interesting! Been thinking a lot on this but also how you can add integrate v0 / lovable as concept designs in issue tracking like linear

this looks like a more structured approach could really help streamline your workflow.

v0 + local repo = design freedom

wait did you not separate backend and frontend before?

nope

can be a pain to remind the agent to keep the business logic out of frontend, but things become so much more stable during fast iterations.

curious how you rate this over figma make

figma make is great i just cant do anything with it when i'm "done" it's just stuck in figma

ah I see. do you know @julianweisser by the way? he has a program you may really like

@julianweisser followed. thanks bub.

I seriously need to start separating my frontend from my backend from my web3 contracts for this reason. 🫠

i usually connect api, backend and integrations in v0 before i bring into cursor on local machine, does alot of the annoying env and stuff but otherwise similar flow

yeah exactly

This is a highly refined and effective workflow. It brilliantly separates rapid ideation from stable development. A masterclass in modern prototyping.

that’s a solid workflow! testing the front end separately sounds really practical for iterating. prototype-first can really save time.

this workflow sounds super efficient fam

真懂产品迭代痛点

great share broo

do i get your flow right? for fast and complete vibe code AI app prototyping 1. get a base from tool that does complete solution fast 2. then continue locally using fake backend first (api plumbing) 3. and then connect to database, hosting, domain if needed