YOUR NEXT EXPERIMENT STARTS HERE
Tell AI what to change. Shape the result on your actual page. Then test it against the original and measure what converts.
No signup. No script to install. Your live site stays unchanged.
Already exploring? Open your workspace ↗
FROM “WHAT IF” TO “WHAT WORKS”
Explore first. Set up and launch when you are ready. Keep the original as your reference at every step.
Enter a public HTTPS URL. Explore your real layout without installing kopair.
Rewrite copy, adjust styles, add a banner or move a block. Refine it in chat or use visual controls.
A is your unchanged original. B is the new variant you shape. Switch views and undo freely.
Sign in, configure goals and your analysis plan, install the SDK, then collect real experiment data.
A PRACTICAL EXPERIMENTATION WORKSPACE
Ask for changes in a conversation, keep the suggestions you want, and tune CSS inside each change. The agent uses the selected element and its style context.
Use Sandbox before installing the script, then edit directly on your connected website in Live. The same selection, chat and change controls work in both.
Toggle individual mutations, compare with the original, undo and redo your work. In your workspace, saved revisions let you restore earlier variant states.
Manage reusable, versioned goals: a destination URL, an element click or a custom event. Choose primary and secondary goals for each experiment.
EVIDENCE BEFORE A DECISION
Set a baseline conversion rate, minimum effect and evaluation period before launch. Follow data collection and evaluate the result with a fixed-horizon statistical test.
Illustration of the setup. No simulated uplift or winning result.
ROOM FOR YOUR NEXT EXPERIMENT
Buy experiment participations when you need them. No monthly commitment. Unused participations never expire.
Loading packs…
When credits run out, experiments pause automatically. A top-up resumes eligible tests; manual pauses stay in place.
USD, taxes excluded. Choose a pack to request a purchase in your workspace. We’ll contact you by email; no automatic charge.
Repeat views in the same test are free. A visitor in three tests uses three participations. Both control and changed variants count. Sandbox and previews are free.
Counted by allocation ID (cookie, user ID or IP). Admission counts even if a change cannot be applied. A pack does not guarantee a conclusive experiment.
LESS IMAGINING. MORE EXPLORING.
Try your idea in a private preview. Your visitors will keep seeing the original site.
FAQ
No. The public demo runs your page in a separate Sandbox. A is the original view and B contains your preview changes. No visitors are assigned, and no experiment or conversion result is created.
Explore a public HTTPS page, select elements, chat with AI, edit styles and compare the original with your new variant. A demo lasts 10 minutes and includes up to 5 AI requests. Changes are temporary; sign in to set up and run an experiment.
Some sites block automated browsers or require authentication. Use a public HTTPS page. If it cannot open, try another page; the demo cannot access private network addresses or signed-in pages.
Sign in, add your site, select goals, prepare variants and set an analysis plan. Install the kopair script and configure your site's connection policy or first-party proxy before launching.
kopair uses a fixed-horizon statistical test with a sample-size plan, confidence intervals and allocation checks. An attractive preview is an idea to test; conversion evidence comes from actual experiment traffic.