PLEN

Product hub

Software that solves a specific problem.

I design, build and run my own products — from the idea, through code and infrastructure, to production. Each one starts with a single, well-defined problem, not with a feature list.

3
products in the making
EU
data and servers in Europe
AI
where it genuinely helps
01 Coming soon

Purvo

SaaS for cleaning companies

Plan the cleaning in the morning. By evening you have photos and the client’s signature.

The problem

Most cleaning companies keep the schedule in Excel, send changes over WhatsApp and have no evidence when a complaint comes in. It works — until a client calls asking whether the team was even there.

The solution

Purvo connects the office and field teams in one workflow. The office plans the week in a calendar, staff receive jobs on their phones, and every job ends with a report: start time and location, before-and-after photos, the client’s signature and any issues raised.

Illustrative preview

What it does

  • Calendar with team colours

    Day and week views, drag-and-drop rescheduling and an instant view of who is free.

  • No app store, no passwords

    Staff get an SMS start code, then sign in with their phone number and a PIN.

  • GPS-verified start

    Location is recorded only at start, finish and when reporting an issue. No all-day tracking.

  • Before and after photos

    Fragile items are photographed before work starts. A job can’t be closed without after photos.

  • Client signature and report

    The client signs off on screen. When they call a week later, you have the answer in a minute.

  • Message translation

    Office messages reach staff in their own language, and replies come back in Polish.

  • QR-code ID badges

    The client scans the badge and checks that the person at the door still works for you.

  • AI planning assistant

    Suggests job assignments based on deadlines and travel time. Never approves anything without a human.

Who it’s for

Cleaning companies — from sole traders to businesses with several teams. Offices, flats, hotels, clinics.

  • Data in the EU (Frankfurt)
  • Roles: owner, manager, team lead, staff
  • Free plan for sole traders
  • 14-day trial of every plan

Technology

  • Nuxt
  • Vue
  • Node.js
  • PostgreSQL
  • PWA
  • Claude
02 Coming soon

apka.space

A platform for AI-built apps

From a conversation to a working app on its own address.

The problem

A small business needs a menu site, a booking system or a simple shop. The options are an expensive agency or a template builder that stops exactly where its real needs begin: its own database, sign-in, payments.

The solution

apka.space doesn’t generate a single page from a template. It runs a full software development process — like a team of people, except the team is made of specialised AI agents. Each owns its stage, and the finished app goes through a quality and security review before it goes live.

Illustrative preview

What it does

  • A team of agents

    Architect, designer, database, backend, frontend, seed data, tester and DevOps — each on its own stage.

  • A real backend

    Apps get their own database, API and authentication, not just static pages.

  • Live preview

    Changes show up instantly in a preview before anything is published.

  • Pre-publish review

    Automated end-to-end tests, a security report and a quality score before every deployment.

  • Custom domain and SSL

    Every app gets an apka.space address or runs on the client’s own domain with a certificate.

  • Payments and invoices

    Stripe and Paynow, invoicing via inFakt, automatic company lookup by tax ID.

  • Marketplace and forking

    Existing projects can be copied and developed further as a starting point for your own app.

  • Code export

    The app’s code belongs to its owner and can be downloaded at any time.

Who it’s for

Small businesses, restaurants, creators and anyone who needs a working app, not just a business card: menus, blogs, portfolios, bookings, shops, dashboards and even games.

  • Claude models (Anthropic)
  • Deployed on Kubernetes
  • Organisations and teams
  • AI usage billed per project

Technology

  • TypeScript
  • Node.js
  • Vue
  • Supabase
  • Kubernetes
  • Claude
03 Coming soon

bets

Football odds analytics

Numbers instead of hunches: where the odds are higher than they should be.

The problem

Online betting tips are usually opinions with no track record. Nobody says how often they were right, or whether the price was even worth taking — and in the long run it’s the price, not the hit rate of a single tip, that decides.

The solution

Every day bets pulls the bookmaker’s offer, builds a detailed statistical profile for each match and runs it through a neural network. The result is compared with vig-free odds from the sharpest market. Every pick is then settled and measured against the closing line — everything is transparent and measurable.

Illustrative preview

What it does

  • Daily data pipeline

    Offer, statistics and odds every morning, with updates through the day — fully automated.

  • Hundreds of features per match

    League tables, team season stats, recent form, head-to-head records and match context.

  • TensorFlow model

    A neural network trained on match history estimates 1 / X / 2 probabilities.

  • Fair odds

    The bookmaker margin is removed from reference odds (devig) to compare true probabilities.

  • Edge indicator

    Offer odds × market probability − 1. A positive edge over the long run is a real result.

  • Settlement and CLV

    Every pick is settled automatically and compared with the closing line — no hiding the misses.

  • Live tracking

    Matches in progress are tracked in real time and results are recorded immediately.

  • Push notifications

    New picks and settlements reach your phone without installing an app.

Who it’s for

People who approach betting analytically and want to see a method and accountable results, not promises. Adults only.

  • Full performance history
  • Model evaluated on historical data
  • Calibration and threshold tuning
  • 18+ only · gamble responsibly

Technology

  • Node.js
  • TensorFlow.js
  • PostgreSQL
  • Nuxt
  • WebSocket
  • Web Push

Approach

The foundation behind every product

The products differ, but the principles they’re built on are the same.

  • Problem before feature

    Every product starts with a specific user situation — a complaint with no evidence, an app a small business can’t afford, a tip with no track record.

  • Own infrastructure in the EU

    I run the servers, databases, certificates and deployments myself. I know where the data is and who can access it.

  • AI that suggests

    AI speeds up the work, but decisions stay with people. No automated scoring of humans.

  • Quality measured, not claimed

    Automated tests, end-to-end tests, monitoring and accountable metrics — even when the result is uncomfortable.

  • Phone first

    Screens designed for the phone first: large buttons, short texts, no app store install needed.

  • Privacy by design

    Each company’s data is isolated, permissions follow roles, and only data that is really needed gets collected.

Technology

The tools I build with

Frontend

  • Vue
  • Nuxt
  • PrimeVue
  • PWA

Backend

  • Node.js
  • TypeScript
  • Express
  • WebSocket

Data

  • PostgreSQL
  • Supabase
  • S3 / MinIO

AI & ML

  • Claude (Anthropic)
  • AI agents
  • TensorFlow.js

Infrastructure

  • Kubernetes
  • nginx
  • Let’s Encrypt
  • PM2

The products are in their final stage of preparation.

Soon each of them will get a link to its public version here. Until then, this page is where all the projects come together.

  • PurvoComing soon
  • apka.spaceComing soon
  • betsComing soon

Want to talk about one of the products or about working together? Get in touch.

grzegorz.sic@scghub.com