Over the years, I conceived products, worked as Product Manager on some, and led teams that built many more.
The examples below are selected from 100+ digital products and solutions created by my teams and me across Deutsche Telekom.
Some started from my own ideas. Some came from business requests that we reframed. Some were products I personally managed. Others were created by teams I had built and led.
Digital Demand Manager — original product posterOKR Board — original product posterMagenta Quiz — original product posterRetro Tool — original product poster
The first question is not:
What should we build?
It is:
What value should be delivered and to whom?
The product comes after that.
People often came to us with a solution already in mind.
“We need Trello.”
“We need a better Excel.”
“Let’s add AI.”
“We need to build our own tool.”
The useful work often started one step earlier:
What change do you need to see happen?
01
Kanban Board
Kanban Board
“We need Trello.”
Teams wanted an agile way to see their work.
We built a lightweight internal Kanban application where teams could create boards, organise work into cards and columns, and adapt the workflow to how they actually operated.
It was simple enough to work for very different teams without becoming another heavyweight project-management system.
Around 100,000 Deutsche Telekom colleagues eventually chose to use it.
A Kanban board was not the goal.
Better overview of progress was.
02
OKR Board
OKR Board
“We need Excel automation for the QBR.”
Teams were maintaining their objectives in spreadsheets and synchronising progress before quarterly reviews.
We built a shared OKR workspace where objectives, ownership and progress could remain visible between QBRs instead of being reconstructed before every meeting.
The product turned a periodic reporting exercise into something teams and leaders could actually work with throughout the quarter.
An OKR Board was not the goal.
Keeping objectives on track was.
03
Excel AI Analyzer
Excel AI Analyzer
“Let’s buy software with AI features.”
Or:
“Give AI chatbots to employees.”
AI was already capable of useful work.
The problem was that much of the business data where it could create real value lived in confidential Excel files — or in systems that could export their data into Excel.
And public AI services were not allowed to touch it.
We built an AI-powered application where employees could upload spreadsheets, ask questions about the data and find patterns in structured and free-text information.
One controlled employee-feedback scenario became Deutsche Telekom’s first approved use of AI on confidential data.
An Excel AI Analyzer was not the goal.
Getting insight from confidential business data was.
04
Refugee Help
Refugee Help
Help starts with knowing what is needed.
People wanted to help, but could not easily see the specific needs of individual refugees.
Refugee Help was designed to make those needs visible.
People could describe what they needed — such as clothes, childcare or an item to use — and specify where and for how long.
The concept connected the intention to help with a concrete request.
A list of donations was not the goal.
Getting the right help to the right person was.
05
Learning Journey
Learning Journey
“We need to share learning material.”
Sharing the material was only the starting point.
People were supposed to move through a learning process, but what they received was essentially a collection of documents.
We turned the material into a digital journey where the sequence, progress and relationship between the different pieces of content became part of the experience.
A Learning Journey was not the goal.
Helping people progress through learning was.
06
Magenta Quiz
Magenta Quiz
“Kahoot is too expensive.”
Kahoot already solved the functional problem.
But the subscriptions were too expensive for the way the teams wanted to use it, and the budget simply was not there.
So we built a reusable internal quiz application that could support different topics, audiences and events without recurring commercial licence costs.
It also gave teams a repeatable way to see how people responded to the content.
A quiz was not the goal.
Measuring engagement was.
More work
More products and prototypes from the same portfolio.
BeHero
A mobile job-discovery concept that starts with employment preferences.
Hiring Buddy
Interview preparation, discussion prompts, notes and candidate evaluation in one workspace.
Event App
A shared event and learning space, shown here in the TDI Classroom experience.
Competence Feedback
A structured workflow for competence evaluation, organisational alignment and feedback.
Agile Performance Evaluation
A digital product for running employee performance-evaluation processes in a more structured and usable way.
MagicBook
An internal knowledge product used by more than 3,000 colleagues, designed to make expertise easier to capture and access.
Procurement & approval solutions
Digital workflows that replaced parts of manual procurement and approval processes, including electronic-signature scenarios.
Local AI chatbot
A locally running LLM interface for scenarios where company information could not leave the controlled environment.
HR AI prototypes
Focused AI products for tasks such as analysis and improving feedback, designed around a specific job rather than generic chat.
SharePoint & Power Platform applications
A broad portfolio of workflows, dashboards, collaboration spaces and business applications built around specific operational problems.
Senso2me
An early-stage product where I designed the interface and built the brand, website and product demonstrations to make a new concept understandable to customers.
MagentaGaming
A B2C gaming product where I was Product Manager and, with design and engineering, changed a development flow in which product changes had taken months into one where user feedback could reach production within days.
The pattern
The products were different.
The pattern was not.
“We need Trello.”
→ What do teams actually need to see?
“We need Excel automation for the QBR.”
→ What needs to stay on track between reviews?
“Let’s buy AI software.”
→ Where is the business data that makes AI valuable?
“People want to help.”
→ How can they see what someone actually needs?
“We need to share learning material.”
→ What experience are people actually supposed to have?
“Kahoot is too expensive.”
→ What value do we need, and what is it worth paying for?
With your permission, I’ll measure which shared link brought you here and what you view or click. No ads or session recording. Withdraw anytime in the footer.
Details
Koloszár András uses Cloudflare to process these optional analytics. A temporary visit ID connects pages, sections and clicks within a visit. Events are kept for up to 31 days, plus limited recovery backups. Your choice is remembered for 180 days.
Withdraw stops analytics and requests deletion of events linked to the deletion keys still held by this tab and other open tabs. Older visits without a remaining key cannot automatically be linked to you. Failed requests can be retried. With daily cleanup, separate deletion-security records remain for up to 32 days from a visit's start, or up to 48 hours after successful deletion; recovery backups are separate. Privacy details
