Kategorien
Allgemein

Das AI Wheel

Nach ein paar grossartigen KI-Diskussionen mit meinen Freund Manu haben wir zusammen das AI-Wheel entwickelt. Hier ein erster Einblick:

Das AI-Wheel ist ein Framework, das Unternehmen dabei unterstützt, künstliche Intelligenz (AI) erfolgreich zu implementieren und alle relevanten Stakeholder einzubeziehen. Besonders geeignet ist es für Unternehmen, die am Anfang ihrer AI-Reise stehen und erste Schritte in Richtung Automatisierung und Effizienzsteigerung machen möchten.

Funktionsweise:

Das AI-Wheel umfasst drei Kernbereiche: Operations, Product und Services. Diese bilden den innersten Ring des AI-Wheels und verdeutlichen, dass das Potenzial von AI ganzheitlich und bereichsübergreifend analysiert werden sollte.

Der zweite Ring des Wheels besteht aus den sogenannten Impact-Sektionen: Accelerate, Evaluate, Transform und Innovate. Diese Sektionen ermöglichen ein strategisches und fokussiertes Vorgehen bei der Potenzialanalyse.

Für jeden Unternehmensbereich wird mit einer Impact-Sections begonnen, wobei sich Accelerate als einfacher Einstieg eignet. Die Teams analysieren gezielt, wo und wie im täglichen Geschäft eine Beschleunigung von Prozessen möglich ist.

Hier bietet der dritte Ring des AI-Wheels Inspiration: Welche Prozesse können beschleunigt werden? Wo verlieren wir unnötig Zeit? Wo lässt sich automatisieren oder die Qualität verbessern?

In ähnlicher Weise arbeitet sich das Team durch die weiteren Sections des AI-Wheels: Zunächst werden die Möglichkeiten von Evaluate und Innovate untersucht, bevor schliesslich in der Sektion Transform mittel- und langfristige Perspektiven und Visionen entwickelt werden.

Dabei wird die Frage gestellt: Wie wird sich unser Geschäftsmodell verändern? Wo sehen wir unser Unternehmen in fünf Jahren?

Je nach Team und Zielsetzung kann es sinnvoll sein, sich zunächst auf bestimmte Bereiche des AI-Wheels zu konzentrieren, z. B. auf Accelerate und Automatisierung, um erste Erfolge zu erzielen.

 

Damit bietet das AI-Wheel Unternehmen einen strukturierten Ansatz, um das Potenzial von künstlicher Intelligenz strategisch zu nutzen. Es fördert eine umfassende Analyse und schrittweise Implementierung, die es Teams ermöglicht, gezielt Fortschritte in Richtung Automatisierung, Effizienzsteigerung und Innovation zu erzielen.

Egal ob erste Schritte oder langfristige Visionen – das AI-Wheel unterstützt dabei, AI nachhaltig in die Unternehmensprozesse zu integrieren und zukunftsfähig aufzustellen.

Kategorien
Innovation myprojects

Exploring the Role of Simulators and AI in Process Understanding

Business and process simulations offer a powerful avenue for dissecting and enhancing both novel and established workflows. In this article, I aim to delve into three pivotal subjects:

  1. 3 Reasons why a Simulator make sense
  2. Harnessing AI for Simulator Development
  3. Training AI through Simulator Integration

Let’s start with the first subject: Why? (part 1)

Reason 1: Fostering Interdepartmental Understanding

In our business, coherence among departments — spanning sales, product, marketing, booking technology, and finance — is paramount. To achieve this, understanding each other’s challenges and perspectives is imperative. Consider the familiar scenario of an IT ticket seemingly taking forever to resolve. Such complexities often elude casual observation, underscoring the necessity for comprehensive understanding across teams.
Simulators are a great way to tackle this issue and create a broader understanding of the process. Colleagues, stakeholders, and even customers can immerse themselves firsthand in the process, gaining valuable insights into its intricacies and challenges.

Example: Navigating the Seat Reservation Challenge

Each spring, my operational team confronts a formidable task: securing seats on our panoramic trains amidst soaring demand.
With most scenic routes already partially booked, the quest for available seats becomes mentally taxing and financially impactful.
If seat reservation is not your daily business, the complexity is hard to understand. However, colleagues in other departments need help grasping the complexity of this issue, and they may need to pay more attention to it when making critical decisions in their department.
Recognizing the need to impart firsthand experience of our seat reservation predicament, I developed a „seat reservation simulator/game“. The physical version of it offers a quick grasp of the process within minutes, making it accessible to all (more about the game below).

Reason 2: Gaining insights while creating the simulator

Beyond merely simulating the experience, the development of a simulator serves as a catalyst for deeper insight into a process’s intricacies.

Two examples:
As I embarked on crafting the seat reservation simulator, I uncovered the significance of seat number structuring and identified enhancement strategies, such as introducing a „W“ designation for window seats. Moreover, it became apparent that team members enjoyed engaging with the simulator collaboratively. This led me to ponder: why not replicate this teamwork dynamic in real-world scenarios by encouraging pairs to tackle tasks together?

Reason 3: Developing a simulator is akin to prototyping

Beyond these insights, the effective deployment of simulators hinges on team involvement during development. This collaborative approach not only offers firsthand experience in prototyping but also underscores the importance of rigorous testing. Indeed, drafting a simulator without iterative feedback, optimizations, and testing is a near-impossible task.

Seat reservation simulator

How the simulator works

Here’s how the seat reservation simulator operates:
Imagine it as a board game, with the board representing a train’s seat map. Specific seats are already occupied or reserved (since we don’t own the coaches and others may have booked earlier).
Players are tasked with selecting eight cards, each depicting a group of passengers vying for seats. Wooden pawns, corresponding to the individuals on the cards, represent them, with each group distinguished by a unique color. The challenge lies in strategically placing these pawns on the board, considering preferences like sitting by the window or facing the direction of travel.
Engaging in this game requires strategic thinking and careful resource allocation. At the end of the game, players tally their points based on successfully allocated seats. I’ve incorporated additional layers and rules to enhance realism, making the experience both challenging and rewarding.

Part 2: Harnessing AI for Simulator Development aka how AI was helping me create the simulator

When I mention „I created“ the simulator, it’s worth noting that AI played a significant role in its development. Initially, I leaned on AI to provide feedback on fundamental game rules. Ensuring clarity in my explanations of the model was a motivating yet challenging aspect of this process.

Utilizing AI to craft the 18 group cards proved to be straightforward. Each drawing, as well as the names of the groups and their originating cities, was generated entirely by AI. Remarkably, this task took less than three hours for all 18 cards—a testament to the efficiency of AI-driven creation. Just envision the time and effort it would have taken to accomplish this manually without the aid of AI.

Part 3: Train the model to play the simulator.

I’m pretty pleased with the outcomes of step 1, „creating a simulator,“ and step 2, „utilizing AI to construct the simulator.“ However, I’m particularly enthusiastic about diving into step 3: training the AI to engage with the simulator actively.
This next phase represents an exciting frontier, and I’ve only begun exploring its possibilities. Rest assured, I’ll provide updates on my progress shortly. Stay tuned for more insights!

Kategorien
myprojects

DIY: Nistkasten mit Livecam

Wir haben von unseren Nachbarn freundlicherweise einen Nistkasten der Vogelwarte Sempach bekommen. Das Ding nur aufzuhängen war mir aber dann noch etwas zu langweilig und ich baute eine Livecam ein.

So klappts:
Das Vogelhaus bzw. der Nistkasten der Vogelwarte eignet sich optimal, da dieser verhältnismässig hoch ist. Das ist relevant, weil die Kamera genügend Distanz zum Boden des Nistkastens braucht, um möglichst das ganze Nest „zu sehen“.

Ich verwende die Home Security Wi-Fi Camera
Tapo C110P4
. Das eine einfache und v.a. sehr preiswerte Kamera (ca. 25.-).

Da ich die Kamera möglichst im Originalzustand lassen wollte, klappe ich den Fuss ganz runter, drehte den Nistkasten auf das Dach und legte die Kamera auf die Decke. Mit einem Post-It erstellte ich eine Schablone für zwei Schrauben links und rechts.

An den Schrauben befestigte ich grosse Unterlagscheiben, welche ich mit Filz ausstattete, um die Kamera nicht zu zerkratzen.

Zur Sicherheit hämmerte ich einen gebogenen Nagel unterhalb der Kamera ins Holz. Diese hat einen Abstand von ein paar Millimeter und sollte ein Runterfallen der Kamera verhindern, falls etwas Unvorhergesehenes passiert.

Anschliessend hängte ich den Nistkasten auf den Balkon und aktivierte den Bewegungs-Alam.

Tatsächlich besuchte bereits zwei Tage später der erste Vogel den Nistkasten, verliess diesen aber wieder. Seitdem steht die „Wohnung“ leer. Bis zum ersten Bewohner werde ich versuchen eine Beleuchtung zu installierten, damit die Kamera nicht auf „Nachtmodus“ umschaltet und nur schwarz-weiss Aufnahmen macht.

Kategorien
Meine Geschichte

Scam 1992

Last Sunday, my neighbour brought an old model railway back on track—the engines and carriages were well stored in a box and wrapped with a newspaper from 1992. Even though I was pretty fascinated by the old models, the paper took my attention even more.

Bill Gates once said, „We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten. Don’t let yourself be lulled into inaction.“

Looking at this 31-year-old paper proves this quote 100%.

Most job ads have a particular gender focus; the ads are very stereotyped, and even though we still have printed newspapers today, the content is very different, and many things feel strange – and somehow wrong.
It will probably be the same if my kids look back to 2023 in 2054, and they will ask us, „How was this even possible?“.

Another fascinating thing is the amount of (printed) scam ads on some pages. These little ads promote „services“ like divorce support, credit help, and of course, massages. The way to contact those shadow companies is to use a chiffre number – the perfect system to stay anonymous – provided by the media company. The scam was probably part of the revenue model.

It was a pleasant experience on that Sunday afternoon, and even though I prefer to look forward, it can be fascinating to quickly look back and see what was „the normal“ just a few years back – and to reflect on how today’s normal will be in 30 years.

Merry Christmas!

Kategorien
Meine Geschichte

Etap 22i

Es hat geklappt: Ich habe eine Segelschiff auf dem Greifensee und zwar ein Etap 22i. Nachdem ich von einem freien Bojenplatz in Niederuster überrascht wurde, wurde ich zum Gebrauchtsboot-Analysten. Mit einer max. zulässigen Länge von 7.5m und einem eigenen Budget von CHF 6000.- schränkte ich mich etwas ein bzw. setzte die Messlatte etwas hoch. Zudem tat ich mich schwer in Genf ein Boot anzuschauen..

Nun habe ich eine Etap 22i i vom Bodensee inkl. so ziemlich alles und einem Anhänger – und für das Alter (1986) ist die „Ente“ gut im Schuss. Keine Osmose, immer mal wieder Teile und laufendes Gut ersetzt, ein paar Dinge extra wie z.B. ein Jütbaum.

Dinge die ich gelernt habe:

  • In der Westschweiz hat es verständlicherweise viel mehr Segelboote zum kauf
  • Inserate mit Vermerken wie „war im Winter immer in der Halle“ relativieren sich bei einem Boot das 30+ jährig ist und gerade mal ein paar wenige Jahre davon beim letzten Besitzer war.
  • Osmose ist ein grosses Thema aber auch ziemlich offensichtlich. Schwierig wird’s beim laufenden Gut und den kleinen Details etc. Ich war hier bereit für CHF 5000 ein entsprechendes Risiko einzugehen.
  • Der „Charakter“ des Vorbesitzer (sofern langjährig) ist wohl der beste Garant für die Qualität. Ein Liebhaber/Handwerker ist m.E. optimal.
  • Die Fotodokumentation der Boote ist meist miserabel. Anschauen ist Pflicht.
  • Gut überlegen, ob es eher ein Grillboot oder eine sportliche Yacht sein soll. Beides ist eher schwierig. Mit der Etap 22i bin ich klar beim Familienboot.
  • Es braucht NFTs für die Bootshistory

Natürlich fängts jetzt erst richtig an mit Unterhalt etc – aber hej der Sommer kann kommen.
A propos Unterhalt: Hier die vollständige Etap22i Dokumentation

Kategorien
Meine Geschichte vr

Import real-world things into VR

My son loves to handcraft trains. His production speed is 2 – 3 models per day, bringing our storage to his limit.
Over the weekend, I was working on a possible solution: a 3D scan of the models.


But not just to solve the storage issue (even with the scan, we will not destroy the trains) but to bring his models to the virtual world and see what happens.
To „scan“ the model, I use the Polycam app. For this use case, it works surprisingly well and much faster than some of the (more) powerful PC applications.
I placed the model into a white styrofoam box and made 250 pictures from random angles. The result is impressive for such a simple process.

The import into the virtual world took me longer as I tried different „worlds“. In my case, Spatial worked the best.
I just had to connect my Oculus to my PC and move the .glb file into the Spatial folder. From there, I imported the object into my Spartial space.

I handed my Oculus over to my son and let him play. His first question was why the train was not moving. I told him I would take this as the next challenge ;)
He enjoyed playing around and scale/turn the object; he probably prefers the real one. However, the digital version may live longer and is a great way to showcase his work to a bigger audience.

This is another example of VR as a powerful „add-on“ to the real world.

But now, I must find out how to make the train move. More to come.

Kategorien
Meine Meinung zu

AI in Travel – an example

AI is rapidly growing in popularity for a variety of tasks, from creating artwork to generating articles. Plattforms like OpenAI provide an innovative and exciting way to explore AI’s potential and its limitations, allowing users to get creative with this powerful technology.

Recently I talked with a friend who works for a global AI company about all the use cases in travel and the idea of using AI for tasks like travel planning.

Can AI replace an agent?

Before I come to this point, here is the OpenAI answer to the question of „how to spend five days in Switzerland“:

There are many ways to spend 5 days in Switzerland, as the country offers a wide variety of activities and sights for visitors. Here is one possible itinerary for a 5-day trip:

Day 1: Arrive in Zurich, take the train to Lucerne, spend the day exploring the city, and take a boat tour of Lake Lucerne. In the evening, take the train back to Zurich and spend the night there.

Day 2: Take a train to the Jungfrau Region, visit the famous Jungfraujoch, which is a mountain pass that offers panoramic views of the Swiss Alps. You can also visit the ice palace and the observatory at the top of the pass, and take a walk on one of the hiking trails. In the evening, take the train back to Zurich and spend the night there.

Day 3: Take a train to the mountain resort of Grindelwald. Spend the day hiking in the surrounding mountains, and take the cable car or train to the summit of the First or Kleine Scheidegg. In the evening, take the train back to Zurich and spend the night there.

Day 4: Take a train to the city of Bern, explore the charming old town, and visit the famous Zytglogge clock tower. In the afternoon, take a train to the town of Interlaken and spend the night there.

Day 5: Take a train to the mountain resort of Wengen, and take the cogwheel train or cable car to the summit of the Jungfraujoch. in the afternoon, take a train back to Zurich, and spend some time shopping and exploring the city before taking the flight back home.

This is just one example of how you can spend 5 days in Switzerland. Depending on your interests, you can also focus on other activities like skiing, snowboarding, paragliding, visiting museums and art galleries, or just enjoy the scenery, relaxing and savor the local cuisine and products. Switzerland is a small country but it’s very diverse, so you can find many options to suit your preferences.

If you know Switzerland, it does not make sense to travel back to Zurich every evening, but otherwise, the result is impressive.

But is it as good as the itinerary from a real travel agent?

Despite the potential of Artificial Intelligence to revolutionize travel planning, few can deny that human touch is still necessary for intricate itineraries and hard-to-find gems. But in the real world, 95% of the agents do not have the capacity to „add“ the human touch. Also, most agents have to known/sell several countries and destinations and nobody can be an expert in 20+ countries.
The 5% are real experts with years of experience and an extraordinary sense for their customers.

However, it is evident that AI significantly impacts travel – and I see great opportunities. Some ideas:

Let’s say we train an AI to create an itinerary with a shallow carbon footprint. Or we train an AI to avoid over-tourism or reduce the risk for a certain type of customer.

I also see an excellent opportunity for the 5% of the experts. Those are the ones finding hidden gems and tracks nobody walked before.

Or in other words: IKEA is great for the mass market and the local carpenter for your unique handmade table.

What do you think?

 

 

 

Kategorien
Meine Geschichte

Retro Joystick DIY

Finally, I finished my retro joystick project – and the result is as beautiful as I thought :)

My daughter helped me to build it using an Arcade Kit from eBay and some spare wood.
The quality of the buttons and the stick is surprisingly good for the price and no driver is needed on Windows 11 (just plug and play). Really cool.

Play

Great project  – and the kids love it as they struggle with the big and ugly state-of-the-art joysticks. Happy days :)

 

Kategorien
Meine Geschichte

Experiencetable (vs. Timetable)

I joined a fascinating and almost philosophical discussion about train/rail travel a few days ago.

In today’s world, train travel is all about a safe & quick journey from A to B. Minimising the time and the number of changes is critical for more than 90% of travellers (and here in Switzerland, the train companies are excellent at delivering this).

Train travel: Faster, faster, faster.

This strong focus on time makes it difficult for leisure companies as their focus is „the experience“ – and not the time.

Considering the small number of leisure travellers, it is evident that leisure is a little mouse next to a cheetah (who loves to run amazingly fast).

Will a cheetah ever go slow?

The question that drove our discussion was whether the mouse is sometimes too focused on the cheetah and should better focus on the neutral ground, aka train tracks.

As shown below, the grid in Europe is incredibly large. Using this grid only to move people & freight as quickly as possible from A to B would be a waste.

Where is the beef? Here.

What would happen if we start to think about tracks as we do about streets, the sea or even the air space? Train tracks are just a grid to enable movement from A to B for whatever reason.

The idea to use the grid to move as fast as possible is established but not so using the grid to have a wonderful experience (don’t blame the cheetah). With this mindset, it is no longer about adding leisure train travel to non-leisure train travel but creating a parallel world.

The „experience-based timetable“.

This is how it works:

Let’s say you must travel from Zürich to Bern; both destinations are little dots in a massive train track grid.

Based on your reason for travelling, you choose the timetable based on time or the timetable based on experience.

Let’s scale this to Europe and the route from Rome to Bruxelles. The timetable based on experience shows you the most beautiful lines, while the timetable based on time shows you the fastest way.

It is up to the reason to travel on which timetable to look at.

This is just an example but with this holistic perspective, we no longer have a pure „product view“ on leisure train travel. This may help to think even bigger – and to may understand the cheetah better :)

What do you think?

Kategorien
vr

Hier ist sie nun: Die Quest Pro

Die VR-Community wartete seit Wochen gespannt auf die Ankündigungen von Meta. Dass am Event eine neue VR-Brille gelauned wird, war schon lange ziemlich klar – und über deren Features wurde wild spekuliert.

Die neue Device bringt so einige an Verbesserungen mit sich. Sehr interessant ist zudem, der Shift zu einer „Hochpreisstrategie“ a la Apple – denn immerhin kostet die Brille knapp $1500.

Die Strategieänderung ist m.E. nachvollziehbar. Einerseits hat sich diese Strategie auch bei iPhone über Jahre bewährt und anderseits fokusiert sich Meta damit auf Kunden, die bereits VR/AR-affin sind. Also weg von der Quantität und hin zur Qualität bzw. zum Upgrade.

M.E. ist eine besser und v.a. bequemere Brille aber auch für die VR-Idee als solches spannend. Die Quest Go und Quest Pro wird in meinem Umfeld oft nach wenigen Minuten als unbequem und umständlich abgestempelt. Das wird bei der Pro sicher anders sein. Oder anders formuliert: Wie lässt sich jemand eher fürs Mountainbiken begeistern? Mit einem 700.- Bike oder einem solches für 4000.-?

So oder so ist klar: Meta ist weit entfernt davon die VR-Pläne auf Eis zu legen.