Last week we attended the World Summit AI in Amsterdam. It seems I'm not the only one excited about AI. It was great to be surrounded by enthusiastic people (who are probably smarter than me). Lots of big names and interesting statistics and discussions at this conference.

Most interesting demo

I think the most interesting demo was Mark Heaps' Groq keynote.
Not only was it well prepared and lightning fast, but it was also well delivered. It was unfortunate that the screen cut out all the time and the sound faltered. This really cast a shadow over an otherwise really good demo.

How AI can be used for sustainability

Decathlon is already deploying AI in several places.

Personalization
80 percent of suggestions on Decathlon's site are personalized.

Product prediction (Forecasting): 1.65 million fewer miles
They have an AI that predicts how many of which products need to be delivered to where. As a result, they have had to drive 1.65 million fewer miles with their trucks since 2023. I thought this was an interesting fact. 

Motto
The motto at Decathlon was “Start small, Scale fast.” 

They also focus on industrializing their AI products to gain the trust of the business. That way, the business automatically becomes an ambassador for their applications. Not coincidentally, this is also part of the vision of The Digitals (parent company of Calibrate). Check out our charter.

Amazonbedrock

Hands on “Building GenAI with AWS Bedrock” workshop

AWS provided a workshop to get acquainted with AWS Bedrock. Before we dove in, some theory and predictions from AWS. 

2025 will be the year of assistants
I think this could be true. 

Fine-tuning and continued training
They similarly explained the difference between finetuning and continued pretraining. 
For finetuning we need a smaller amount of labeled data, and for continued pre-training we need large amounts of unlabeled data. 

AWS APP builder
There was also mention of AWS APP builder, this would be an application to quickly create ai powered applications without knowledge of code. 

 

AI on ubuntu

Industry experts meet and network

It was also an excellent opportunity to meet the right people from Microsoft, Canonical (Andreea Munteanu), Sopra Steria, EY (Raf Ganseman). 

We also had a nice restaurant visit and evening activity with the people from 

  • Vectr.Consulting (Jony Van Puymbroeck & Tom Franckx)
  • Tensr (Bruno De Deken)
  • Cronos.ai (Fiore Frequelli and Bart Briers)

 This is how I got to know Tom https://indy-me.ai/

Volvo efficiency
Attending the AI summit
Pepsico embracing AI
FBI and AI
Room Yadda Yaddda
room panel
European flag on slide
FBI slide
FBI slide 2
FBI slide 3
Huggingface slide
Huggingface slide 2
Huggingface slide 3
FBI slide
NES GPT slide
Decathlon slide

World Summit AI 2024 (Amsterdam)

FBI slide

AI safety and law

Did you know that Voice data is also PII?

There were several speakers or panels from legislators from Europe, but also from the United States (FBI / Homeland security). Impressive to see at what level they are working on it there.  

NesGPT

AI for productivity - Case from Nestlé - NesGPT

That RAG provides a productivity boost I didn't have to tell you. How nice that Nestlé came to bring a wonderful case on this. They initially started with a simple RAG, to prove that it could work. Soon they switched to multi agent RAG. They are building their RAG there with Langchain and on Azure AI search. 

Importance of metadata
What we already knew was also confirmed by them. It is super important to determine what metadata you put along with each chunk (this is something we can do nicely in Drupal with the search_api connector in an interface). This is key to the success of the “search”. 

Plugging in other tools
What they call Function calling at OpenAI, they call “AI connectors” at Nestlé. 

Scalable
They quickly scaled from 100,000 to 200,000 users. 

Man In The Middle
This is normally the name of a hacking attack. They also developed a Man in the Middle themselves. A kind of cookie notification before their AI communicates to the outside world: 

“Do you really want to share this question, or the information from this conversation, with the Internet?”. 
Very cool, as I had seen a similar implementation on questions.ai. My personal prediction: we are going to see this more and more, this will pretty much become the cookie notification of the AI. 

Quality control
Nice also to hear that they really need to do quality control, because when switching a model they don't want to have to retest everything. 

The value of AI in contact centers

There were also several sessions where met told about the use of AI in contact centers.  Here I heard that “Voice data is PII.” But the value of RAG and connectors on Genesys were also discussed here.

AI at work

If you want to see an AI at work be sure to check out https://vlaanderen.vragen.ai. 

Practical issues

There were a lot of practical issues during the event, though. 

  • Presentations (on main stage) dropping out
  • Audio issues (to put it mildly)
  • Timing issues (substandard time management in the various halls)

Whether I will go again next year is still to be seen. 
I was at Drupalcon a few weeks ago and now I appreciate so much how smoothly everything went there. Could it have something to do with the fact that it's Open Source events run by volunteers?

Closing notes

Anyway, it was worthwhile to be there. I was able to take away interesting figures and see great cases. Send me a message if you want to know more about this!

Would you like to chat about this? Give me a call!