Custom apps for companies

Custom software, connected to the systems you already use

Instead of spreadsheets, copying and retyping data: a working app in a few weeks that reads and writes data in your ERP, CRM or online store. AI where it actually helps. The code and accounts stay with you.

  • 9 years of building business apps in .NET
  • Integrations via API, import files and databases
  • A prototype before you order, then a fixed price
Order queue3 messages
  • Metalplast DemoZK/2026/10/0412 · 5 itemsready
  • Agro-Serwis DemoOrder_AG-118.pdf · 6 items2 to confirm
  • Termo-Linia DemoRESENT: TL/552/2026duplicate?
From the documentPU hose 8x5.5 blue – 2 rolls
Our item codePN-WP-08-NB2 rolls × 50 m = 100 m
confirm conversion
Sample app with fictional data

When an app makes sense

Sound familiar?

Companies have an ERP, a CRM or an online store, and yet part of the work still happens on the side: in spreadsheets, e-mails and manual retyping. That usually means a small app or a connection between systems is missing.

  • Orders, jobs or requests live in an Excel file that travels by e-mail in several versions.
  • Nobody knows who changed what or which version is current.
  • Some of the data has to be retyped into the ERP anyway.
  • Approvals happen by e-mail or by word of mouth.

A dashboard with roles, approvals and change history, connected to your ERP, replaces the spreadsheet without replacing the whole system.

  • You enter the same data in the store, the ERP and the CRM.
  • Export from one program, fix it in Excel, import into another.
  • Stock levels and prices drift apart between systems.
  • The report for management is assembled by hand from several sources.

An integration via API or import files moves the data on its own, with an error log and an alert when something doesn't match.

  • Someone retypes orders and requests from e-mails and PDFs line by line.
  • Customers use their own codes, “2 rolls” instead of metres or “3 packs” instead of pieces.
  • An address or serial number is missing, so you have to ask again.
  • The same e-mail arrives a second time, with extra lines added.

Here AI reads the document, the app checks the data against your system, and an employee approves it. There's an example in the demo below.

How I work

From a conversation to a working app

First we check whether it's worth it. Then a prototype, a fixed price and a build in short stages.

  1. 01

    Call

    You show me what you do by hand and which programs you use. I tell you whether an app makes sense.

    30 minutes, free
  2. 02

    Prototype

    You see the app working on your own examples before you decide to build it.

    usually 1–2 weeks
  3. 03

    Scope and price

    We write down what the app should do, what it connects to and how you will accept it. The decision is yours.

    scope · fixed price · deadline
  4. 04

    Build

    A working version to review every week. Tests and the connection to your systems.

    paid in stages
  5. 05

    Launch and maintenance

    Deployment, documentation, then monitoring, backups and fixes on a monthly plan.

    code and accounts stay with you

What you get

Fast thanks to AI, solid like a developer's work

AI helps me write code faster, so you pay for fewer hours. Below is what separates an app for everyday work from a prototype put together in one evening.

Connected to your systems

The app reads and writes data in your ERP, CRM, store or database via API or import files. No manual copying.

store→app→ERPno retyping

Roles and approvals

Everyone sees what they should. Important operations need approval, and the change history shows who did what.

manager→approved 14:32

Tests

The key paths have automated tests, and acceptance is done on your own examples.

Data security

Passwords and keys outside the code, backups, access only for authorised people. GDPR rules agreed before we start.

AI under control

AI where it helps, e.g. reading e-mails and PDFs. A person approves the result.

Code and accounts stay with you

The repository, server and documentation are on your accounts from day one. After payment I transfer the rights to the code written in the project. You can hand maintenance to anyone you like.

Manual fallback and monitoring

If something stops working, you know right away and can work the old way until I fix it.

Example

Order queue from e-mail

The app reads orders from e-mails and PDFs, checks them against the catalogue, and an employee approves them before they are imported into the ERP. Three cases on fictional data; you can click every decision.

  • Source preview for every line
  • Unit conversions and item code selection
  • Duplicate detection and a ready-made message to the customer

A video walkthrough of the demo is coming soon.

What I can build

Examples of apps and integrations

I start every project with one specific piece of manual work. If an off-the-shelf product or a feature of your ERP is enough, I'll say so.

  • Connecting two systemsERP with your store, CRM or marketplace tools: data flows on its own, with no exports or copying.
  • A dashboard instead of a spreadsheetShared data, roles, approvals and change history instead of file versions circulating by e-mail.
  • AI document readingOrders, requests and enquiries from e-mails and PDFs go into the system after approval.
  • Reports from ERP dataDashboards and summaries your software doesn't provide, calculated automatically instead of in Excel.

Principles

How we work together

  • What you already have comes firstI check the features of your software and existing products. If they are enough, I say so.
  • A prototype before you orderYou see a working prototype on your own examples before ordering the whole thing.
  • A fixed price for a written scopeScope changes are quoted separately. You pay in stages, for what already works.
  • A working version every weekYou see progress and can adjust direction along the way, not only at the end.
  • Data under controlFirst tests on anonymised examples. Access to real data only after the rules are agreed in writing.
  • Everything on your accountsCode, accounts and documentation stay with you. You can end the cooperation after any stage.

Stages and pricing

We start with a call, you pay for a specific stage

Each stage has a clear scope. After each one you decide whether we continue.

This is where we start

Call

free
30 minutes, online
  • You show me what you do by hand or in a spreadsheet
  • I tell you whether an app makes sense
  • Even when an off-the-shelf product will do

Prototype

$380net
up to 10 h, 1–2 weeks
  • A working prototype on your examples
  • A written scope for the whole app
  • A fixed price and deadline for the build

The amount is credited towards a build ordered within 90 days.

Build

from $1,530net
fixed price for a written scope
  • Paid in stages
  • A working version every week
  • Tests, deployment and documentation

7 days for acceptance and a 90-day bug warranty. A small app or integration usually takes 3–6 weeks.

Maintenance

from $130net / month
or 2% of the build price
  • Monitoring and backups
  • Bug fixes and updates
  • Response: critical within 1 business day, others within 3 days

The higher amount applies; the minimum fee covers up to 3 h of work per month. Further development is quoted separately.

Licences, servers and AI model costs are covered by the client and agreed before we start. Net prices (+ VAT where applicable), valid from 4 October 2026. Amounts in USD are converted from my PLN prices at 1 USD = 3.91 PLN (NBP, 5 October 2026). Process reviews and AI consulting: on request.

About me

A developer who asks about the process first

For 9 years I have been building business applications in .NET: processes with roles, approvals and change history, and integrations between systems. I have worked on systems for sales teams, on a professional exam and certification system, and on an HR platform.

I built the AI app SwipeKing.ai on my own: from comparing models for quality and cost to the infrastructure. I use AI to write code, but I read and test every change. I also know where AI helps inside an app and where a plain rule in code works better.

  1. since 2021Sticos · Senior .NET DeveloperHR platform: absences and health & safety training, REST API in .NET, Angular.
  2. 2025–2026, in parallelSwipeKing.ai · my own AI projectAll of the tech: model selection, backend, frontend, infrastructure.
  3. 2018–2021DXC Technology · .NET DeveloperProfessional exam system: results, reports, certificates; microservices and SOAP integrations.
  4. 2017–2018eLeader · Junior .NET DeveloperApp for sales representatives: products, reports, users.
  5. 2016Comarch · internshipReporting tool for Comarch ERP XL and Altum.

MSc in Computer Science, Lublin University of Technology.

FAQ

Frequently asked questions

Does this replace our ERP or CRM?

No. The app works with the programs you already use: it reads data from them and writes results back via API or import files. Accounting and warehouse documents are still created in your system.

Which companies is this for?

Companies that have an ERP, CRM, online store or job management software, and still work alongside it with spreadsheets, e-mails and manual retyping. The industry matters less than how much of that manual work there is.

Why not an off-the-shelf product or a no-code tool?

They are often enough, and then I'll tell you. A custom app makes sense when ready-made software doesn't support the way you work, and a no-code tool can't connect to your ERP or meet your requirements for permissions and data.

Code written with AI. Will it still work in a year?

AI speeds up writing, but I read and test every change and take responsibility for the code as if it were my own. The app has tests and documentation and lives on your accounts, so another developer can take it over. I don't paste your customers' data into AI tools.

What if the AI in the app makes a mistake?

Where AI reads documents, every value shows its source and uncertain items are flagged. Nothing goes into the system without an employee's approval.

What about our data and GDPR?

We start with anonymised examples. Before I get access to real data, we agree the rules in writing: the parties' roles, the scope of data, service providers, retention time and deletion when the work ends.

How long does it take?

A prototype usually takes 1–2 weeks. A small app or integration is usually 3–6 weeks; I give the exact deadline together with the fixed price after the prototype.

Who will maintain it?

I will, on a monthly plan: monitoring, backups, fixes and updates within an agreed number of hours. I respond to critical issues within 1 business day and to others within 3 days. The code, accounts and documentation are yours, so you can also hand maintenance over to your own IT person.

What does the contract look like?

Before signing, you get a draft contract to read. The build price is fixed for the written scope, you pay in stages, you have 7 days for acceptance and a 90-day bug warranty. After full payment I transfer the rights to the code written in the project. General modules I reuse across projects (e.g. login) come with a perpetual licence including the right to modify them, and their code is in your repository too.

What does day-to-day cooperation look like?

I work remotely, in English or Polish. Calls and support during agreed hours: Mon–Fri 9:00–12:00 (Polish time, CET/CEST). I don't offer 24/7 on-call support, which is why every solution has a manual fallback.

Contact

Show me what you do by hand today

In two sentences, tell me what you keep in a spreadsheet or retype between programs, and which software you use. I'll reply with a proposed time for a 30-minute call.

CallsMon–Fri 9:00–12:00 (Polish time, CET/CEST)
What is it about?
Two sentences are enough. Please don't paste customer data.