Meet the Mida Agent — now on every account. It's an AI teammate that operates Mida from chat: ask about results across all your live tests, have it create or edit experiments, fix a variant, or write and rate hypotheses. It remembers you across conversations, reads files you upload, and hands off to support when it's stuck.
What it can do:
Build and run tests — Create experiments with variants, goals, targeting, and schedules. Start, pause, conclude, or reopen a test, and roll a winner out to all traffic.
Read results — Tells you whether a test is ready to call, using Mida's two checks: the chance of beating control must clear the test's confidence bar, and expected loss must be under 10%. Breaks results down by device, country, or traffic source, and checks whether a test's data can be trusted.
Suggest what to test next — Starts from your own list of test ideas and past wins and losses, then compares against what's worked for other Mida customers. Reads your live pages first, so suggestions are grounded in the copy that's actually there.
Fix and extend variants — Finds the right elements on the page, repairs a variant that isn't changing anything, and adds or removes variants on an existing test.
Keep an eye on things for you — Runs checks on a schedule, emails you when a test is ready to call, and writes PDF or CSV reports for clients.
Personalization — Builds campaigns for named companies or job roles, or matches your landing pages to your ad campaigns.
To access it:
In the left sidebar, click Agent.
Type your query into the chat box (e.g., "What should I test next on my site?") and press Enter.