Paraxis Labs from Wrocław builds its own digital products and custom software.

ReactTypeScriptPythonFlaskPostgreSQLpgvectorRAGOCRAWSTerraform

We build software that turns hours of tedious work and searching into productive time. We test solutions on our own products and then deliver them to clients, fitted to how they really work.

From chaos to a working system

We replace spreadsheets, emails, paper notes and old software with one system. Your team works in one place and has what it needs at hand.

SpreadsheetsEmails and phone callsPaper formsLegacy software

Team dashboardAutomatic quotesLive reportsCustomer app

Work and products

Systems we build for clients and products we develop ourselves.

ITS System

A sales and production system for a vehicle body manufacturer: configurator with pricing, quotes and a shop-floor panel.

Client project · in progressSee the project

Modiva Flow

A new product we are working on right now. More soon.

Product · in developmentDetails coming soon

Edu-verse

AI that grades handwritten solutions like an examiner and justifies every point.

Product · liveSee the product

Why us

Four things we can show rather than just promise.

We start from your process, not from technology.

First we look at how your team works and where time gets lost. Only then do we choose the solution.

You pay per stage.

Every stage has a scope, a price and a written sign-off.

AI where it pays off.

We use AI when it genuinely cuts work, not for show. Everywhere else, good code is enough.

You talk to the people who write the code.

One team from analysis to maintenance, with no middlemen between you and the developers.

How we work

A project with a defined scope

We split the project into stages. Each has its own quote, schedule and sign-off. We start the next stage only after the previous one is accepted.

YouParaxis

You describe the needA talk about what is not working
An ideaA problem to solve
Waiting
Analysis and specificationProcesses, documents, scope
Specification
Waiting
You approve the scopeStages with price and deadline
Stage plan
Waiting
We build the stageA working part of the system
Stage 1
Waiting
You test itOn your own data
Your dataFeedback
Waiting
Fixes and releaseWithin the stage scope
Production version
Waiting
Written sign-offStage closed
Acceptance report
Waiting

An example project in three stages. We set the number of stages after the analysis.

Let's talk about your project.

Briefly describe what you need. We'll get back to you and set up a call.

contact@paraxislabs.comWrocław, Poland

A website, configurator and panel where a vehicle body manufacturer runs sales and shop-floor work.

Client: ITS System sp. z o.o., Siechnice, Poland
Industry: Type-approved vehicle bodies
Our role: Analysis, design, build and maintenance
Status: Rolled out in stages

ITS builds truck bodies on Mercedes-Benz, Iveco, Renault Trucks, Ford and MAN chassis. It started with a new website and a configurator. The analysis showed that the company lost the most time in its processes: quotes were made in spreadsheets and shop-floor hours were logged on paper.

So the website grew into a system in which ITS runs part of its operations. We deliver every module separately, with its own sign-off.

Preparing a quote

1 h10 min

Average time to prepare one quote, before and after launch.

Labour time standards moved from paper into the system

104

Standards list from June 2026.

Chassis versions from five brands in one configurator

Ford, Mercedes-Benz, Renault Trucks, Iveco and MAN, each with its own GVW and body lengths.

Problems we solved

Pricing without spreadsheets

Problem: The price and weight of a body depend on the chassis, length and equipment. They were calculated in a spreadsheet, by hand, for every configuration.
What we did: A configurator with a pricing engine in which the rules are data, not code. It detects configurations that cannot be built, such as an exceeded gross vehicle weight or a length that does not fit the wheelbase, and redraws the dimension drawing live.
Result: Nobody can configure a vehicle that cannot be built, and the weight and price are ready at once. Quotes are generated automatically.

The dealer picks a chassis.

The configurator replaces spreadsheets and email threads.

Adds a body.

Body type, length and equipment come from a list matched to the chassis.

The system prices and weighs it.

Live, following the rules stored in the configurator.

A PDF quote goes to the customer.

In one click, with a saved quote history.

Illustrative diagram of the configurator.

One quote, three views

Problem: The same configurator serves a customer, a dealer and an ITS salesperson. Each of them should see different information.
What we did: Three visibility layers enforced on the server. Every role gets exactly the data meant for it, and the rest never leaves the system.
Result: An amount never reaches the browser of someone who should not see it. Dedicated tests guard this.

ChassisGVW up to 3.5 tBodytarpaulin dropside, 4.2 mEquipment3 options

Customer

Dealer

ITS salesperson

One configuration, three visibility layers. Data hidden in this mock-up.

Quotes and sales in one place

Problem: Quotes, website enquiries and customer data had no common place, so it was hard to see all sales at once.
What we did: Saved quotes with continuous numbering and frozen pricing, a print-ready A4 document, quote statuses, a website enquiry inbox, a customer register with duplicate detection, salesperson results and a dealer module.
Result: Sales and management work on the same data, and every quote has its own history.

ITS panel: list of quotes with number, configuration, author and status (amounts blurred)

Continuous quote numberingAmounts frozen on the issue date

Shop-floor hours

Problem: Working time on the shop floor was logged on paper, against a printed list of time standards.
What we did: A shop-floor panel where workers log time by standard code, several entries a day, up to 14 days back. The manager sees the hours and absence balance, and management sees the statistics.
Result: Shop-floor hours are in the system the same day, with no retyping of paper slips.

What I did todayTotal for the day6.5 h

Activity code100

Hours1.5

Sample activity · standard 2 h

Save

Logged today

100Sample activity A2 h

210Sample activity B3 h

330Sample activity C1.5 h

ManagerWeekly balance

Worked38 h

Absences1 day

ManagementStatistics

Match with standards94%

Orders in progress12

Mock-up on sample data.

Data in management’s hands

Problem: Every change to an option price or a time standard would have needed a developer.
What we did: A data management panel with validation and a dry run before saving. Plus bug reporting from every screen, with a screenshot and technical context attached.
Result: Management updates standards and configurator items on their own, and reporting a bug takes a moment.

ITS panel: bug report window that automatically attaches the screen address, browser and recent errors

Bug report from any screenTechnical context attached automatically

Technology

Next.js, React, TypeScriptStatically prerendered website for search engines, with the configurator and panel in one app.
Flask and PostgreSQLCatalogue, pricing engine and price visibility layers on the server side.
Permissions as dataRoles and module access are described in a dictionary. A new role needs no code changes.
SecurityPermissions checked on the server for every request, and a separate account for every user.

Do you have a similar process built on spreadsheets and paper?

Let's talk

NextEdu-verse

From chaos to a working system