← All projects
PL/EN
Full case study · 10 min · June 2026
ClientPirxey
Tech stack
ActivePirxey · 2026AI-native software house, Gdańsk
01 / case study

Pirxey stopped fitting
the company
to the tool.

Pirxey, a company with 70+ engineers, built a custom internal operating system. Time tracking as the first module, built in ~20 working days, AI‑native from day one. The foundation for the company’s full OS.

70+engineers
Company size
AI-native
dev, one
pair of hands
 
ClientCustom company software
Pirxey

AI-native software house from Gdańsk, 70+ engineers, digital products since 2018. pirxey-os is their internal operating system, used company-wide since July 2026.

pirxey.com
169 active accountsCustom company softwareAI-nativeMCP
Fig. 1 — tracker · schedule · calendarpirxey-os v1.0
169accounts
Active users
One system for the whole company.
AI-native
dev, one
pair of hands
Working model
No software house. From architecture to production deployment.
1 module
Just the beginning
Time tracking as the first module of the company OS.
Full
company
rollout
Since July 2026
The whole company logs time in pirxey-os. Clockify is switched off.
Showcasepirxey‑os — product presentation in a real flowRotato · 41s

For years, custom software was reserved for the largest companies. Expensive, slow to deploy, risky. That’s why small and medium companies chose SaaS tools. Even if they didn’t fit perfectly, they were faster and simpler to get started.

Pirxey started asking a different question:

Łukasz Gralińskifounder & CEO · Pirxey

“Can AI actually help build real, working software today:a system that can handle a company’s daily process?”

That’s how pirxey-os was born. From day one it was about something bigger than replacing a single tool.

Thesis

Custom software is no longer a luxury. Pirxey stopped fitting the company to the tool and started building a tool fitted to the company.

SaaS works fast at the start.
Then the company starts
fitting itself to the tool.

“I have to do it differently than I need to. This doesn’t work the way our company works.”

Missing features were only part of the problem. Companies rarely have just one SaaS. They start with one tool, add more, glue them together with integrations, until the process is scattered across integrations, exports and spreadsheets.

DataAverage number of apps per companyCompanies of all sizes
Companies under 50 people
36
Companies 50–150 people
65
Companies 150–1000 people
93
Enterprise (1000+ people)
94+
Source: Okta Business at Work 2024.

Time tracking as the first module
of the company’s internal operating system.

AI has changed the economics of building software. A custom system can be delivered faster and cheaper than a few years ago. The conclusion is practical:

If a process is critical, repeatable and expensive in SaaS tools, it’s worth calculating whether it should become its own system.

Module 01 · now
Time Tracking
Module 02 · done
Reports
Module 03 · Q3
Profiles
Module 04 · Q4
Calendar
Module 05 · Q4
Finance
Module 06 · 2026
AI Agents
Module 07 · 2026
Mobile

Diagram — from “Clockify replacement” to company OS, one development path

Why build

Six reasons why a custom system beats the next SaaS

From fit, through AI, to the math.

  • 01FitThe system follows Pirxey's process.
  • 02ConsolidationReplacing several SaaS tools stitched together with integrations.
  • 03Ownership, no lock-inThe software stays in the company as its own asset.
  • 04Grows with the firmEvery new need becomes the next module.
  • 05AI-agent readyA UI for people, API and MCP for agents.
  • 06The math worksBuild vs buy: a one-time build and an asset the company owns.
01/ 06

Software that fits the company.

SaaS tools are built for hundreds of thousands of companies at once. Pirxey was one of them, and its process had to adapt to the tool.

pirxey-os works exactly the way Pirxey works. The lower cost comes along with it.

Fig. 3Tool-to-process alignmentSaaS vs pirxey-os
Generic SaaS
Company fits itself
to the tool
workarounds in sheetsempty, unused fields
vs
pirxey-os
Tool fitted
to the company
no workarounds100% coverage
02/ 06

Many processes in one system.

Companies rarely have just one problem to solve. They start with one SaaS, add more, glue them with integrations, until the result is a side effect of decisions made separately.

pirxey-os is one platform. One data model, one permission layer, one place for every process.

TToggl
JJira
HHarvest
CClockify
FFloat
BBambooHR
pirxey-os
One coherent system
Time trackingReportsProfilesCalendarFinanceAI

Diagram — ~36–93 apps in a typical company (Okta, 2024) → one coherent system

03/ 06

Ownership and no vendor lock-in.

If a process is critical to the company, it’s worth asking: do we want to rent it from a vendor, or own it as a company asset?

A subscription is rent. After four years, all you’re left with is paid invoices and nothing more. A custom system after that same time is a company asset that keeps growing.

SaaS · rent
You pay again every year.
Y1
Y2
Y3
Y4
The same cost comes back every year. After four years all you have is paid invoices and nothing more.
pirxey-os · ownership
One-time build, the system stays as an asset.
Y1
Y2
Y3
Y4
One-time build, and the system stays as a company asset that grows in value.
04/ 06

The system grows with the company.

The biggest advantage of custom software is that it grows with the company — new processes become new modules.

Every new need becomes the next module of the same system.

Time tracking
Now
Next step
Time tracking
Next step
Next module
Previous
Time tracking
Roadmap
AI Agents
More modules
Previous
Time tracking
Horizon

Visualisation — every new need is the next module of the same system

05/ 06

Open to AI agents via API and MCP.

pirxey-os exposes data through a public REST API and MCP server, so AI agents can operate the system alongside people using the interface.

pirxey-bot
online
Today
Logged. Pirxey Dashboard · “Code review” · 1:30 (today 14:21). You’ve logged 6h 15m today.14:21
Thanks. Can you summarize my week?14:22
  • Pirxey Dashboard — 18h 10m
  • Med4You — 9h 30m
  • Internal — 3h 20m
  • Meetings — 1h 40m
14:22
Write a message…

The same API and MCP work underneath, used by other agents too

The same endpoint that answers questions in Telegram can be called by an n8n workflow, a Claude agent, or a CI script.

REST APIMCP serverSupabase RLS

More on this model: BR-Budget case study — a finance app where the primary user is an AI agent.

// pirxey-os as an agent tool
{
  "tool": "pirxey_log_time",
  "project": "Pirxey Dashboard",
  "task": "Code review",
  "duration_min": 90,
  "date": "2025-06-20"
}
06/ 06

And the numbers add up faster
than you’d expect.

Comparing the annual SaaS cost to the one-time build cost of pirxey-os.

SaaS · annual cost
~162k
PLN/year · 169 accounts × ~80 PLN/mo.
  • Recurring cost every year
  • Roadmap owned by the vendor
  • Limited control over the process
  • After a year: only paid invoices
pirxey-os · company asset
~50k
one-time build · ~20 working days · rough estimate
  • A system the company owns
  • Own roadmap and future modules
  • Full control over data and process
  • After a year: an asset that keeps growing
Year 1 difference
~112,000 PLN
That’s what stays in the company after twelve months, along with an asset that keeps growing.

The goal here is a company asset. The saving comes with it.

This is pirxey-os
in production.

Custom design system built for Pirxey: with colors, components and light and dark mode. Below: some real screens — tracker, schedule, calendar and reports. The same system, different work contexts.

Tech stack

TypeScriptReact 19Next.js 15SupabasePostgreSQLMCPTailwind CSSREST APIRow-Level SecurityVercel
If we know exactly how Pirxey works, why should we fit ourselves to a generic tool? We wanted to see whether we could build a system that works our way: first for time tracking, and later as the foundation for the company’s other processes.
Łukasz Gralińskifounder & CEO · Pirxey · On the pirxey-os decision

Delivered as a full product —
from architecture to deployment.

This is real software handling a real process in a real company: architecture, application, API, historical data import, security, tests and production deployment. All in one pair of hands, without a software house.

  • 01Discovery & architectureUnderstanding how time tracking actually works in a company with 70+ engineers.
  • 02System foundationFoundation designed from day one for the full multi-module OS.
  • 03Design systemCustom color system, components, light and dark mode.
  • 04API & AI integrationsPublic REST API, MCP server, layer for AI agents.
  • 05Security & deploymentPermissions, RLS, Sentry, tests, production environments.

Since July 2026 the whole company has been logging time in pirxey-os. The rollout is done and Clockify has been switched off.

Planned next: more processes as modules of the same system, including people, tech profiles, estimates, quotes and tickets.

One system.
One database.
A tool fitted
to the company.

pirxey-os replaced Clockify and gives the company one platform for its processes. The system fits itself to the company.

And it stays in the company as an asset.

169
Active accounts, company-wide
1 module
Just the beginning
~50k PLN
Build cost

Interested in your own system?

If you’re paying for SaaS that forces workarounds, let’s check whether a custom module would be cheaper and better fitted to your process.

Business-first CAIO · designer · developer. I build systems from idea to deployment: solo, with AI as the tool.