The question "where do I start automating my business" usually comes down not to technology but to a fear that it will take months, require a developer, and eat a big budget. In practice, you can get your first working result in a week — no consultants, no rewriting your processes, no drawn-out integration projects. Below is an honest seven-day plan: what to do on each day, why the order matters, and exactly what result should be in place by the end of the week. No promises of a "revolution" — just a sequence of steps you can realistically walk through on your own.
Why the first step is your inbox, not a CRM
Most automation advice starts with rolling out a CRM or some "single source of truth." The problem is that the data for those systems still has to be typed in by hand, while the actual work of the business keeps living somewhere else — in email. That's where supplier invoices, bills, certificates, price lists, contracts, and deadline reminders land. Your inbox isn't just a communication channel; it's the de facto archive of your company's operational life — already full, and already organized by sender and date.
So the first step in automation isn't to build a new system from scratch — it's to connect to what you already have. If you've accumulated emails with attachments over several years, they already hold the entire history of your supplier relationships: who, when, at what price, and on what terms. All you have to do is let the system read them over IMAP — nothing to migrate, nothing to move by hand. We covered how supplier emails get lost and what that costs you in our piece on lost supplier emails.
Day 1: connect your inbox over IMAP
On day one there's a single task — give HORUVIA access to your mailbox over IMAP. It's a standard protocol supported by every email service; connecting it comes down to entering an address and an app password. Nothing gets deleted or forwarded anywhere: the system reads what has already arrived and works with a copy. There's no code to write — this is exactly the kind of automation without a developer that people usually only talk about.
As soon as you connect, an onboarding preset kicks in for your segment. The presets are built in advance around typical scenarios: import operations, retail on marketplaces, services, hospitality, logistics. A preset is the answer to "but how do I set this up for my case": it suggests which document types to expect, which deadlines to track, and which alerts to switch on, so you're not configuring everything from a blank page. You pick your profile — and the system already understands the context it will be working in.
Days 2–3: the system sorts your inbox and builds registers
On the second and third days the real work begins — but it isn't yours, it's the system's. HORUVIA sorts your incoming mail by supplier, brand, and document type: which is an invoice, which a price list, which a certificate, which a contract, which is just correspondence. Scans and photos of documents are run through OCR so that the text becomes searchable and parseable, rather than staying a "picture."
Out of that sorting a supplier memory is built — a structured base where, for each counterparty, you can see their documents, their price history, and their certificates with expiry dates. In scenarios with a large catalog, that memory can cover thousands of product lines across dozens of brands — and it isn't an export from someone else's system, it's assembled from the mail itself. You don't have to enter anything: the registers are built from what has already arrived. That's the essence of the "small-business automation, step by step" approach — each new layer of data emerges from the previous one instead of being created as a separate project.
It's worth understanding the difference between this and a classic CRM: a CRM expects you to fill in the records, whereas a system that lives in your inbox fills them in for you. We drew out that contrast in more detail in AI operator vs. CRM — in short, the difference comes down to who does the data entry.
Days 4–5: switch on alerts and the deadline engine
By midweek the data is sorted, and you can turn on the whole reason for doing this — so the system doesn't just store information but tells you about what matters. On the fourth and fifth days, Telegram alerts and the deadline engine come online.
Alerts cover situations you used to catch by chance, or miss entirely: an invoice arrives, an offer comes in with a deadline, a certificate is about to expire, an email has gone unanswered for too long. There's also the "went down → came back up" principle: if a process failed, you get not only a signal about the problem but a paired signal that everything has recovered — so you're never left in the dark.
The deadline engine takes on the dates that are easy to miss: contracts, declarations, insurance, health certificates, shelf-life dates. Reminders arrive early — weeks and then days before the deadline — so you have time to react rather than learning about a lapse after the fact. Setting up these reminders is part of the preset, so by day five you're not assembling rules by hand — you're just confirming what to track.
What you'll see by the end of the week: your first invoice as a payment draft
The high point of the week is your first invoice turned into a payment draft. The system parses the document, pulls out the details, amounts, and payment purpose, and prepares the payment as a ready-made draft. And here's a point worth saying plainly: it isn't artificial intelligence that moves the money. The calculation and the payment itself are handled by deterministic code — predictable logic, no "creative" model behavior — and the final decision is always made by a person. The draft is ready, but you're the one who clicks the button.
That order removes the biggest fear about automating finances: the system saves you time on routine data entry, but never gains the right to move money. By the end of the week you don't have an abstract "system that's been implemented" — you have a concrete result: a sorted inbox, live registers, working alerts, and a first payment ready for you to approve.
Why a week, not a month
A one-week timeline is realistic precisely because none of the steps require rebuilding the business. You don't change your processes — you connect to what's already happening. Your inbox is already full, the documents have already arrived, your suppliers are already writing to you. The system reads that stream, sorts it, and starts prompting you — instead of forcing you to reshape your work around it.
That's exactly why the first step in automation should come from the data side, not the interface side. Once the data is sorted, everything else — alerts, deadlines, a Kanban task board, a Company Pulse showing shipments by stage, order recommendations by seasonality and ABC analysis — is layered on top without separate projects. The week goes into the foundation; from there, features get added as you need them.
But what about data security
A fair question when you connect your inbox is what happens to the data. Two principles matter here. First: the system works with a copy and deletes nothing — in line with the general rule of "only add, never delete." Second: sensitive operations, above all payments, stay with deterministic code and a human, not with the model. We covered separately how the data is handled and why the AI layer is kept apart from the money circuit in our piece on data security when working with AI.
Understanding those boundaries removes most of the anxiety: automation doesn't mean "hand control to a robot." It means taking manual sorting, hunting through attachments, and chasing deadlines off people's plates — the work that eats hours and still gets done with mistakes.
Where you should start
If it all comes down to a single action: start by connecting your inbox and choosing a preset for your segment. From there the system itself will show you what in your flow needs attention — and you'll add features not from a checklist in an article, but from what actually surfaces in your own mail. That's the honest path: not "implement everything at once," but see your first result in a week and build from there.
Want to gauge how much time and money this sorting would return in your specific case — run the numbers on the calculator, or book a demo where we connect your inbox and show you your first parsed invoice on your own data. You can also check whether the approach fits your segment through the FAQ — but the best way to grasp the value of automation is to see your first payment draft, assembled from your own inbox.