How our lovable AI Agent saved Davos WEF (and booked us 17 CXO meetings) ????

This is a story of how we created an AI that:

1. Solved a real life problem of messy logistics in Davos (and will solve it again in 01/2026)
2. Became a universally loved, go-to AI solution for hundreds of users
3. Impressed the right people with our tech and booked 17 meetings with CXOs in 4 days

tl;dr ???

• We built an AI using our own tech that guided 600+ Davos attendees to the best events, restaurants, helped with sprained ankles, buying groceries, parking...?
• Unlike Salesforce, our solution didn’t need a mobile app + had more users/usage + was more inclusive.?
• People loved our Davos AI so much that it scored us both inbound and outbound meetings with our ICPs
• we also had a killer time meeting the CEO of Uber , CPO of OpenAI , CEO of 高盛 , Chief Scientist at Meta and others... but we left that for the very end of this post.

what's the story?

so... here's the problem ?.

“Davos is easy…” Said no one.?

Well, maybe it is, if you're a high ranking executive with a supporting team of EAs scheduling your agenda.

Everybody else? Faces the same organizational/logistical "Herculean task" ???.

Everybody else has to sign up, RSVP and register - just like you would at any other conference.

Thousands of executives, hundreds of events. Even the most connected attendees struggle to keep up.

But first? You have to find the events. There's a couple hundred events to review.

They're everywhere:

• Luma
• Eventbrite
• WhatsApp groups
• Websites with varying structures and legibility

A mess, from a data standpoint.

so... here's the solution ?.

This year, after 5 “Davoses” of organizational chaos - we decided to fix that.?

Using Quickchat AI , we built an AI that became the go-to tool for navigating the World Economic Forum in Davos.

• It answered event questions, found private dinners, and even gave restaurant and transportation tips - all in natural conversation.
• By the end of the conference, 600+ high-profile attendees were using it.?
• We booked 17 meetings on-site.?
• Influencers attending Davos with 600k+ followers shared it on Instagram, organically.

And we didn’t spend a dime on paid marketing.?

I mean… that’s not entirely true.

For fun, we did slap QR codes on our car, and mounted 2 flags.

Say what you want... people took pictures, scanned the QR code and recognized Quickchat AI.

so... how did we do it?

Step 1: Data Gathering ??? ?? ???

We knew that navigating Davos, a sprawling conference with 5000+ attendees, hundreds of events, and tight schedules, was a headache, even for the best-prepared.

So, we rolled up our sleeves and spent 30+ hours researching everything:

• Event names, times, and locations (with Google Maps links)
• Speaker bios
• Badge requirements for private and semi-private events
• Registration links for last-minute RSVPs
• Transportation tips and more

We ended up with a 400+ event spreadsheet - over 5000 cells of data!

(...to be honest, this was the most laborious part - just goes to show that anyone wanting to do AI has to have a reliable, standardized and clean dataset)

?? Wanna see the spreadsheet? It's here ??

Step 2: Data ?? AI ?? ?? ??

We wrote a script to turn our monster ?? Excel sheet into an AI-friendly text format.

It followed the same structure for each event, making it reliable for the AI to provide relevant answers.

We added it to our AI's Knowledge Base in the Quickchat AI platform and started testing.

Step 3: Making the AI lovable ?? ???? ??

Using our platform, Quickchat AI, we built an intuitive AI designed to handle all the complexity of Davos information retrieval.

The bot worked like this:

1. Ask it anything: “What are today’s top events?”, “Where is the 西门子 dinner?”, or even “What’s the best way to get from Congress Center to Belvédère Hotel?”
2. Instant, precise answers: The bot pulled data directly from the structured event knowledge base we built, giving tailored responses instead of generic ones.
3. Natural conversation: It felt like chatting with your friend who’s been to Davos 8 times.

The AI wasn’t limited to text fields on a webpage. We integrated it with WhatsApp, making it easy for users to text the bot like a personal assistant. 35% of users used it that way. If they preferred a more formal interface, they could still access it through the landing page.

This dual-access approach helped ensure the tool was available to anyone, no matter their preferred method of interaction. We could embed the AI into both platforms while customizing it with our branding for a professional, polished finish.

(You can still see the landing page here: quickchat.ai/davos. The event’s over, but it gives you an idea.)

Step 4: “Beta” Testers ???

We shared the AI link with closed WhatsApp Davos groups from previous editions. Why?

• To gather feedback before the big launch
• To refine the Knowledge Base based on real questions

Within one weekend, 260 high-profile users exchanged over 700 messages with the bot. They asked about events, venues, and logistics, helping us polish the experience.

We added every missing bit of information to the Knowledge Base and worked on the tone.?

Step 5: The Big Launch ??? ?? ??

By Sunday, we were ready.

The first helicopters and private jets started flying into Davos.

The Launch? ??

• We shared our AI in every Davos-related group and social media thread we could find.
• We optimized the landing page for mobile because most users would check it on their phones.
• Most importantly, we enabled a WhatsApp integration - so our AI became a contact in their phones, providing the easiest access possible.

The Impact? ??

• 600+ users out of 3000 attendees. A massive number, considering that this is a conference where most attendees have human assistants managing their schedules.
• 17 meetings. We booked face-to-face time with top executives.
• Unprecedented visibility. We’re still sorting through follow-ups from people who loved the tool.

The Takeaway? ??

If you’re a startup and want to grow, there’s probably something (a process, a workflow, “things that have always been done this way”) you can do 10x better using AI.

Another example? ??

I really liked the recent Notion Faces campaign, but was surprised that I actually need to create my “face” manually (I guess we expect AI everywhere at this point).

Lasqo AI , a startup building a tool to generate on-brand visual assets, felt the same way and created a tool where you simply:

1. Upload your portrait
2. Get a custom Notion-like portrait in a few seconds (shoutout to Ferdinand Terme ??)

We’ll see more and more case studies like this. If you can use your own product to do this, BIG bonus points.

Davos Tax - Dara Khosrowshahi , Kevin Weil , Yann LeCun , David Solomon , Demis Hassabis , @david rubenstein (no LinkedIn), Patrice Louvet , Justin B. Smith - to name a few!

See you next year, folks!