New campaigns open at Amazon's suggested bid
Ask the agent for a campaign and every target starts at the bid Amazon recommends for it, with the range shown, instead of a flat default. Name a bid yourself and it uses that instead.
What's new
New things in the Autron agent and the ad engine, written when they land. Once a month the best of them go out as one email from Julian.
The story so far
You talk to your account and it does the work: it answers from your own data, runs the recurring jobs as routines, reaches you on email, Slack and WhatsApp, and bids toward one goal per product while standing back wherever you step in.
Ask the agent for a campaign and every target starts at the bid Amazon recommends for it, with the range shown, instead of a flat default. Name a bid yourself and it uses that instead.
Where you've stepped in, the engine stands back. Goals now shows every exception in one place: what you changed, whether it helped, and when it hands back to Autron.

Every conversation now knows today's date, your marketplace's clock and the Amazon events ahead. Ask it to hold spend back until a date and it sets a ceiling on each target with an exact end, then hands back.
Pick any model. Sonnet 5.5 replaces Sonnet 5, and GPT-6.1 Sol is now the default on paid plans; Luna stays the free one. Change it from the model picker in any chat.
Link your number in Settings › Integrations. Routines deliver there, you can send a photo of a listing, and the conversation carries on in the app. Nothing changes without your approval.
The agent that runs in the app now answers on the website too, from the help centre, with no account needed. Press ⌘I on any page, or the Ask agent button on a phone.
Every article rewritten against the current product and published at autron.ai/help. The agent answers from the same pages, so what it quotes is what the page says.
"Put everything in the kitchen line on 30% ACoS" is one request, one approval and one job, however many products it covers. Each product still gets its own goal.