Explore more ideas. Decide in seconds.
Unship turns one round of chat into a handful of real, rendered directions in your running app. Flip through them, keep the best one, and your agent cleans up the rest.
See how it works.
Your agent writes a few real directions into your code. You keep one. The rest never ship.
Click ‹ › or press ← → to flip directions.The page swaps instantly: real code, not previews.
Grab the bar and move it anywhere.A dashed ghost shows which of six spots it'll snap to.
Double-click to tuck the bar into a little dot.Click the dot to bring it back.
More groups on the page? The bar grows a group row.Click it, pick a group, and the same bar drives that part of the page.↑ ↓ cycles groups.
Hold ~1s to keep. It copies an instruction for your agent.Paste it, and the other options are removed from your code.
setup
Ask for design options. Compare them in your local preview. Pick a direction in chat, and your agent cleans up the unused variants before shipping.
faq
When should I reach for Unship?
When chat is getting too abstract and you need to look at the options in the app. Ask for a few directions, flip through them, keep the one with taste, and let the rest disappear.
What should I ask my agent for?
Use normal language: compare 4 hero directions, explore empty and error states, try 3 CTA rows, render a few README openings, or show me different button system treatments. If you can judge it by seeing it, it is probably fair game.
Do I have to wire up the variants?
No. Your agent writes the temporary options in source and mounts the local picker when the app needs it. You stay in reviewer mode.
What does "keep" actually do?
Holding the pill copies a keep instruction for your agent. Paste it back, and the agent removes the losing options so the winner is just normal app code.
Does this become a production experiment?
No. Unship is for local taste decisions before you ship. It is not analytics, targeting, persistence, or an A/B platform.