Business process automation: invoices, quotes, stock and attendance.

02 / 08

Invoicing and incoming invoices

Done by hand, invoicing means that at the end of the month someone opens technicians' timesheets, delivery notes or partner reports, adds them up in Excel and issues invoices one by one. After automation, invoices are calculated straight from the data, fixed rules check them before they go out, for example whether the price matches the contract, and they reach the accounting software without retyping.

Supplier invoices arrive by email as PDFs, receipts come in an envelope at the end of the month, and someone types them into POHODA, Money or KROS. After automation, a language model reads the document, rules check the totals, VAT and duplicates, and the data goes into the program through an import. A person only sees the documents the system isn't sure about.

For an ordinary volume, an off-the-shelf tool such as Doklado or Digitoo is often enough, and if that's your case I'll tell you. When your own solution pays off, for example with foreign invoices or your own checking rules, is covered on the page about invoice data extraction.

03 / 08

Inquiries, quotes and contracts

An inquiry comes in through a form, an email or a phone call and stays in the inbox of whoever took it. When that person is on holiday, the customer waits. After automation, every inquiry lands in one list with a named owner, and if nobody responds within the agreed time, the system alerts the team lead. If customers mainly want an appointment, for example for an inspection or a service visit, a booking system connected to a calendar helps. I built one for this website, connected to Google Calendar and Google Meet.

Salespeople often put a quote together by copying an old one. They change the name, the address and the prices and hope the previous customer isn't still in there somewhere. After automation, they pick the customer and the items, the system fills in the company details by its registration number and the prices from the price list, and creates a PDF from the template you keep in Word or Google Docs. A contract is created the same way once the customer accepts the quote.

In a service company that might be a repair quote based on the labour and parts price list, at a wholesaler a price list for a new customer with their discount, and in manufacturing a contract for work with delivery dates. How I generate, sign and archive them is on the page about contracts and quotes from templates.

04 / 08

Orders and stock

At wholesalers, orders from customers often arrive as a PDF or as text in an email. Someone enters them into the program, checks whether the goods are in stock and sends a confirmation. After automation, the order is read, compared with stock and the customer's price list, and a draft document appears in the program for a person to confirm. Orders from the online store you already have can flow into stock and accounting on their own.

When data moves between systems, it matters which system is the source of truth and what happens when a transfer fails. I log every transfer, failed ones are retried, and an error triggers an alert. The page on connecting systems covers this in detail, including the routes into POHODA, Money, KROS and ABRA Flexi.

Stock in Excel works as long as one person edits it. Once the storekeeper books goods in, a salesperson books them out and a temp does the stocktake, the numbers drift apart. After automation, goods in and out are recorded on a phone by scanning barcodes right at the shelf, and everyone sees the same stock levels. In manufacturing, material consumption per job can be tied to stock as well. When this grows beyond a single process, it becomes a custom ERP with stock, production and attendance.

05 / 08

Attendance, deadlines and onboarding clients or employees

Employees often record attendance in a notebook or a spreadsheet, and at the end of the month someone retypes it for payroll. After automation, they clock in and out on their phone, the manager approves leave in the app and the payroll data is ready as an export. I build attendance as a standalone app or as an ERP module.

Deadlines live in individual people's heads and calendars: a contract running out, a service check on a customer's machine, an invoice falling due, documents a client still hasn't sent. After automation, the system tracks them and sends a reminder to the customer or a task to a colleague in good time. I work this way myself in Pult, my own CRM, which pulls in Gmail, Google Calendar and Slack, tracks deals, tasks and invoices, and which I use daily. I build the same thing for accounting firms that spend their days chasing documents from clients.

When a new client or employee starts, the same details often get entered in several places: the CRM, the accounting software, the contract, a shared folder and an email with logins. After automation, you enter them once, for a company the registration number is enough, and the system prepares the contract, creates the records and sends a welcome email. Decisions that belong to a person stay with that person.

06 / 08

Reports and dashboards for management

A weekly report is often made by someone downloading exports from three systems on Monday morning, merging them in Excel and emailing them to management. By the time anyone opens it, the numbers are old. After automation, the dashboard fills itself from the accounting software, the stock system, the CRM or carrier data and shows what management actually follows, such as revenue by customer, unpaid invoices, open jobs and stock below minimum.

In logistics these are typically carrier reports, in manufacturing order progress, in a service business open jobs, and in an accounting firm the clients who are late with their documents. A simple automation is enough for a morning summary by email or in Slack. A dashboard with filters, where everyone sees only their own numbers, needs a small app with logins.

07 / 08

When n8n, Make or Zapier is enough and when you need custom code

Plenty of processes don't need programming. n8n, Make and Zapier have ready-made connectors for Gmail, Outlook, Google Sheets, Slack and hundreds of other apps, and I can set up a simple workflow in them faster and cheaper than in custom code. n8n can also run on your own server, while Make and Zapier run in their vendors' cloud and the subscription price grows with the number of steps your automations run each month.

Custom code pays off when the rules have many exceptions, when the accounting software runs on your own computer or server (POHODA through mServer, for example), when the data volumes are large, or when people need a screen where they review and fix exceptions. I often combine both: simple steps stay in n8n and the complex part is written in code.

I only add AI where text without a fixed structure has to be read, such as a customer email, an order in a PDF or a scanned document. A language model reads and suggests, fixed rules check that the numbers add up, and a person decides the unclear cases. I process data in the EU wherever possible and call models through business APIs, which under the providers' terms are not used for training by default. More on this on the page about AI agents for business.

An off-the-shelf tool without custom code is usually enough when:

• The workflow has fixed steps, for example a filled-in form becomes a spreadsheet row and an email to a colleague
• You're connecting common cloud apps that the tool already has connectors for
• You're dealing with tens or hundreds of records a month
• Nobody needs their own screen for approvals or corrections
• The data arrives in a fixed format, for example a bank or online store export, so no AI is needed

08 / 08

What process automation costs and how we work together

Automating one process usually takes a few days of work and starts at €500. If several automations grow into a system the company runs on every day, such as a CRM, an ERP or a client portal, I price it by scope and in parts. Running costs, meaning hosting, backups and any AI calls, come to tens or a low few hundred euros a month depending on volume.

The consultation is free. After the first call you get a specific price and date, not a range. I split the quote into parts, so you can start with one process and add the next once the first one is running. Once you agree to the quote or some of its parts, I'll build a free prototype on your data so you can check that it will work before you make the final decision.

You pay 20% of the price when you order and the rest on handover. By default I run the automation on my own accounts for a monthly fee that covers hosting, backups and updates, and I always fix bugs in it free of charge. If it works with personal data, we sign a data processing agreement under Article 28 of the GDPR before launch.

To start, tell me in the form below which process costs you the most time and which programs you use for it.