Data entry from documents
Orders, supplier quotes and delivery notes arrive as PDFs or scans. AI extracts the data and enters it into your system, and a person only checks and confirms.
- PDFs and scans
- Into your ERP
- Human check
AI in business
AI that does the work, not just writes emails
ProCoding, from Split, Croatia (EU), brings AI into business processes: data entry from documents, sorting enquiries and emails, answers from your own knowledge base. Connected to your systems, inside your own Azure subscription, with the saving measured in euros before anything goes further.
Vlado Pandžić · Founder · Senior .NET architect
14 years building business systems, and connecting AI to them through Azure OpenAI. Microsoft Certified: Azure AI Fundamentals. You talk to him directly.
var order = await chat.GetResponseAsync<PurchaseOrder>( $"Extract the order from this email:{email.Body}"); var check = rules.Validate(order.Result); if (check.IsValid) await erp.CreateDraftOrderAsync(order.Result);else await review.QueueAsync(email, check.Errors);- 20%
- of EU companies use AI (Eurostat 2025)
- 71%
- don’t adopt it for lack of expertise
- 14
- years building business systems
- EU
- company, contract and invoicing
Where AI saves the most
Work with lots of repetition, text and clear rules. That is where the saving is measured in hours a day, not minutes.
Sorting enquiries and emails
Enquiries from the website, email and forms arrive sorted by type and urgency, and go to the right person, instead of someone reading and forwarding them by hand.
- Emails
- Forms
- Routing
Answers from your knowledge base
Staff ask a question and the answer comes from your own guides, contracts and procedures, with a link to the source so you can see where it came from.
- Your documents
- Source with every answer
- Internal assistant
Draft replies to customers
A draft reply or quote based on your price list, terms and customer history. A salesperson checks it and sends it in a minute instead of half an hour.
- Price list
- Terms
- Draft reply
Questions over your data
"How much of product X did we sell in Q3?" An answer from your own database, without waiting for a report or exporting to Excel.
- Database
- Reports
- Summaries
AI inside your application
Have your own .NET system? We add AI directly to it, where people already work, instead of making them switch to a separate window.
- .NET
- Azure OpenAI
- Your subscription
Three ways to start
Each has a clear outcome, so you know what you get before we begin.
Why AI stalls in companies
Rarely because of the technology. Almost always it is one of these five.
Everyone uses ChatGPT, nobody knows for what
Everyone in their own way, with no rules. The saving is minutes per person, and company data ends up in public tools.
It started with a tool, not a task
Licences were bought and everyone waits for someone to find a use. The real start is the task that eats the most hours, with the tool chosen for it.
The demo works, your documents don’t
Everything works on three examples from a presentation. Your documents have exceptions, poor scans and special cases, which is why the pilot runs on real data.
AI answers, and someone still retypes it
Without a connection to your systems, the saving disappears into copying from one window to another.
Nobody knows whether it saves anything
Without measuring before and after there is no decision on the next step. So we count time, accuracy and cost per case.
How we bring in AI
One process after another, because only the numbers tell you what comes next.
- 01
One process
A concrete task with a lot of manual work and a clear measure, not "AI across the company".
- 02
Rules
What may leave the company and what stays. Data stays in your own Azure subscription.
- 03
Pilot
On your real data, with a person approving every result.
- 04
Measure
Time saved, accuracy and cost. Only when it pays off does it go into daily work.
When we are a good fit, and when not
We are a good fit when
- +You have work that repeats hundreds of times a month: documents, enquiries, emails
- +People spend part of the day retyping, sorting or looking for data
- +You want company data to stay with you, not in public tools
- +You use Microsoft 365, Azure or your own .NET systems
We are not a good fit when
- −You want a chatbot on your website just to have one
- −The task happens a few times a month, so development would never pay back
- −You expect AI to work without any human check from day one
- −You need your own AI model built from scratch
Tools
- Azure OpenAI
- Azure AI Document Intelligence
- Azure AI Search
- Microsoft.Extensions.AI
- .NET
- Azure Functions
- SQL Server
- Entra ID
Questions
With one task that involves a lot of manual work and has a clear measure, such as entering data from documents or sorting enquiries. Not with a tool, and not with the whole company at once. Once the first process shows numbers, choosing the next one is easy.
It depends on how many cases you have and how long one takes. For example, 1,500 documents a month at six minutes of manual work each comes to a saving of about €29,000 a year. We do the calculation with your numbers before a euro is spent on development.
No. We use Azure OpenAI inside your own Azure subscription. The data stays under your control and Microsoft does not use it to train models.
That is why a person approves every result until the numbers show AI is accurate enough, and the system checks the rules you already know. Where every mistake is expensive, AI only prepares and a person decides.
AI takes over retyping, sorting and looking up data. People check the results and do the work that needs judgement, and the company can grow without hiring more people for the same work.
Since February 2025, companies that use AI must make sure the people working with it know what they are doing. That is why clear rules and short staff training are part of every rollout.
The assessment is agreed after a free 30-minute call, and the pilot has a fixed price. The AI itself is paid per use, usually a few cents per document or enquiry.
Using AI only for emails? Let’s see where it would really save you money.
In a free 30-minute call we go through your processes and see where AI would save the most, and how much.
Book a free 30-minute call →- EU company, Split, Croatia
- Data stays in your subscription
- NDA on request
- You talk to the engineer who writes the code
Useful to include
- /Which task takes the most manual work
- /Roughly how many cases a month
- /Which systems you use (ERP, CRM, email)
- /Whether you use Microsoft 365 or Azure