Who to call today.
A premium cured meats producer, the food manufacturer in our other case studies, lives on customers who order again and again. We built them one system: a daily list of who to call, an AI assistant the sales reps talk to, orders read from photos, sales brought in from their old software, and a catalog made from one product list.
Sources: the project's database (data loaded April to July 2026) and its code, read 21 October 2026.
Every morning, a list of who to call and why.
Each customer orders at their own pace: one every week, another every month. The system learns that pace from past orders, so it knows who is due, who is late and which call is worth the most today.
Move the days forward and watch the list change.
The calls most likely to bring an order today go first.
"Per call" is the customer's usual order. The later they are, the more it counts, up to three times.
An AI assistant the sales reps just talk to.
Reps ask by voice, text or photo: who to call, what a customer usually buys, log this sale. Amounts and dates come straight from the system. The AI repeats them, it never works them out itself.
The AI suggests, the rep decides.
The system itself refuses, whoever asks, the AI included.
A photo of an order becomes an order, ready to check.
A customer sends a handwritten list, a WhatsApp screenshot or a PDF. The AI matches each line to the company's product list and fills in the order. What it can't read stays blank, it never guesses a price, and the rep checks before saving.
Order · Thursday
Product A — 12 kg
Product B — 6 kg
Prod∿∿∿ — ∿ kg
Product C — 8 kg
Deliver Monday
- #04Product A12 kg
- #17Product B6 kg
- —Not readable, left blank?
- #09Product C8 kg
Sales from the old software, without retyping.
The company's software can't connect to anything: it only exports reports. The system reads each report and adds up every invoice. If its sum doesn't match the report's total, it stops and saves nothing, so a wrong number never gets in.
The totals match, so everything is saved.
Diagram with example figures. This step runs in the background: there is no screen for it.
order lines brought in from those 183 invoices, dated 24 March to 13 July 2026, without anyone retyping them.
One product list. The catalog builds itself.
The company keeps one list of its products. From it come the web catalog, a page to share with buyers, the PDF and the label sheets. Add or remove a product and the PDF rebuilds itself, contents and page numbers included.
The catalog we made for this client.
Real project · brand and contacts removed20 of its 24 pages, in Portuguese, as the client uses it. We removed the logo, the brand name, the website and the product codes. The 4 pages left out name the company or its people.
What this means for your company.
You may not sell cured meats. But if your customers order again and again, the same pieces fit your business.
Your past orders turn into a daily list: who is due, who is late and which call is worth the most. No customer goes quiet without anyone noticing.
See the CRM → Chapters 02 and 03An assistant for your sales teamReps ask by voice or text, log sales on the road and turn order photos into orders. Clear rules decide what the AI may change.
See the AI sales assistant → Chapter 04Use the software you already haveIf your system only exports reports, we bring them in and check every total, so you don't have to replace it or retype anything.
See the operational dashboard → Chapter 05A catalog that stays up to dateKeep one product list. The PDF, the web catalog and the label sheets update from it, with no designer redoing pages.
See the product catalog →Tell us how your customers order today.
Tell us your problem →Every demo on this page uses example data, except the catalog in chapter 05, which is the real one. Client name withheld.