Ten feature blocks
Order rhythm including trend, basket size, price band, strong categories with a half-life, discount profile, returns, viewed but not bought, and open baskets - all derived and recalculable at any time.
Instead of a campaign calendar there is an agent: it derives a persona from every customer's behaviour, checks fifteen possible occasions and decides per customer whether there is a reason to write at all. Below a minimum score nothing is sent - better nothing than just something.
A classic newsletter goes to everyone at the same time with the same content. This agent turns that around: for every single customer it asks whether there is an occasion right now, picks the strongest one and writes the message for it. Without an occasion the inbox stays empty.
Order rhythm including trend, basket size, price band, strong categories with a half-life, discount profile, returns, viewed but not bought, and open baskets - all derived and recalculable at any time.
Replenishment for consumables, abandoned basket, reactivation, matching add-on after a purchase, repeated interest, birthday, new arrivals in a strong category, unredeemed voucher and more.
Every occasion gets a score. If the best one stays below your threshold, nothing is sent. It is the single most important setting and the reason frequency stays low by itself.
Every message lands as a draft in the approval queue - with reasoning, score, text and HTML preview and a subject line check. From there: accept, reject or send as a test mail.
The autopilot only kicks in once an occasion has collected enough manually approved messages. Automatically sent ones do not count, otherwise the threshold would be worthless after the first run.
There is no synchronisation with an outside platform. Only anonymised facts go to the model - never names, never addresses.
Newsletters rarely lack reach - they lack an occasion. Writing less often but at the right moment keeps your list alive.
For consumable articles the agent knows the typical interval and gets in touch shortly before the product runs out - not three weeks early, and not once the customer has already bought elsewhere.
Because decisions are made per customer, large segments are not needed. Even with a few hundred recipients you get meaningful communication instead of a blanket mail.
There is no editorial plan somebody has to maintain. The occasion comes from the data, the copy from your template and writing rules.
The agent proposes, you decide - and afterwards you see what worked. The reporting goes well beyond open rates.
Mail template and writing rules define how the copy sounds and looks. Your brand stays recognisable even though the wording is produced per occasion.
Length, promotional trigger words, exclamation marks, percentages, capitals and emojis are checked and scored - that costs nothing and keeps subject lines out of spam suspicion.
Per newsletter you see which product source produced the revenue - due consumables or filler. That is the real lever, not the open rate.
For any customer you can view all occasions side by side with score, reasoning and verdict. The most common support question is answered in seconds.
Reach funnel, revenue classes, discount profile, range breadth, rhythm trend, order days, returns and the "valuable and at risk" list.
Personas, group counts, metrics and reporting can be limited to one sales channel - otherwise the numbers add up across all your brands.
Own tables, own sending path, no external platform. Installed by upload in the admin, without SSH.
Personas are calculated in the shop and stored there. Only anonymised features and aggregate numbers go to the language model - no names, no addresses, no order history in clear text.
You store your own API key. Costs are settled directly with the provider, nothing is routed through our servers.
Own SMTP configuration per sales channel, sender freely configurable. Reporting covers opens, clicks, orders and revenue with last-touch attribution in a configurable window.
Building the personas, running the agent and maintaining the groups can also be triggered from the console - handy for the initial build and for cron operation.
How the agent decides and what happens to the data.
Instead of a campaign calendar the agent checks per customer whether there is an occasion right now. It derives a persona from behaviour, scores fifteen possible occasions and only writes when the best one reaches a minimum score.
Rather fewer. Without a sufficient occasion nothing is sent, and you set the threshold. The principle is: better nothing than just something.
Buying directly costs 49.99 euros net per month, 249.99 euros net per half year or 449.99 euros net per year. On top come your AI provider costs. All prices are net and exclude VAT.
No. Personas are calculated in the shop and stored there. Only anonymised features and aggregate values go to the language model, never names, addresses or order histories in clear text.
Not at first. Every message lands as a draft in the approval queue. The autopilot only starts once an occasion has collected enough manually approved messages, and it can be switched off entirely.
Two routes, the same feature set - pick whichever suits your setup.
All prices net, plus VAT.
You buy the licence directly from us and receive an invoice from us. Installation by upload in the admin without SSH, updates through our licence platform, support straight from the developer.
The plugin will be listed in the official Shopware Community Store. Installation and automatic updates then run through the plugin manager in the backend, billing through your Shopware account. We will link it here as soon as it is available.