Introduction

Well, welcome everybody.

Happy that you are all here.

I'm talking about how to start with AI on a personal level and also

on a company -wide level.

Before I start I want to maybe, guys, can you raise the hand if you have ever been before here at this beautiful campus?

You can see it back there, a lot of people, amazing, cool.

Cool.

About Merantix and Its Pillars

Merantix Overview

For those who haven't been here before, maybe some few words about Merantix.

Actually, Merantix, it's not only Merantix.

We are like four divided in four pillars, right?

Pillar 1: Merantix Capital

The first one is Merantix Capital.

These are like the investment guys investing in all these startups, mostly pre -seed.

Pillar 2: Merantix Momentum

Then we have Merantix Momentum, what is the consultancy part where I'm working on.

and also yeah a bunch of other people you already know.

We are having like a

Research and Strategy Teams

small research team.

We are doing like really AI research lately on like small

LLMs.

We have like a strategy team but mainly we have like a large

delivery team that is really building AI solutions.

Pillar 3: Merantix AI Campus

The next one is the Merontics

AI campus as you can see here.

We have like three floors.

This is like the main

area the heart where you are and this is where the magic happens we want to bring

people together like events like this one we more or less have like 100 events

of this per year and yeah we want to create like an environment where we can

connect to each other right we have like innovation hubs from big corporates we

have these startups what Melanthix Capital is investing in they can chat

with us with our research team and so on so we really want to create this vibrant

vibrant community here.

Pillar 4: German AI Association and AI House

And the last one is the German AI Association and AI House in Davos

for the World Economic Forum.

That's more or less our gate to the public sector.

Speaker Introduction

Just briefly introduce myself.

I'm Ferdinand Schwarz.

I'm working at Merantix Momentum.

We

are sitting just around the corner over there.

And I'm mainly working in the public sector

as an innovation partner where we guide the management but also teams from

ideation, validation, prioritization of AI use cases up to the kickoff and also

towards like scaling these solutions.

Navigating the AI Tsunami: Four Phases

I want to start with a picture and maybe

it feels familiar for somebody of you with all these news AI tools right we're

living in such a rapid time, it's even like increasing, it somehow feels like to get a bit

lost, right?

So my goal today is to give you like a glimpse idea of how to navigate through this AI

tsunami, what is coming up front.

And I want to do it to you and explain how to do this in four

different steps.

Phase 1: The Baby — Play and Explore

And that's adapted to the human life cycle.

So we started with phase one, with the

the baby.

When you observe babies, they are being born in the world, they don't know how

to act on it, and they start playing around with everything.

Do not even know what they

are doing, but just playing around and try to somehow get connected and get at some idea

of the world.

Basically, that's also my or our advice when to start with AI.

You see

all these fancy names, fancy tools like Gemini, OpenAI, LangDoc and so on.

You don't need to have like a master degree or be like the best programmer in the world

right now, right?

For these tools, it's just starting, go on these websites and build in 10 to 30 minutes

your maybe like own app, right, with Lowable or your own website.

You can start cloning your voice with 11 labs or create media photos with Black Forest labs.

Automate things like Stefano showed you guys.

It's like you just go there, watch a YouTube video, and then you can basically taste and get a gut feeling of how these AI tools are working.

Phase 2: The Student — Educate Yourself

The next phase, as you remember, is being a student in school, university.

And there, humans or we learned in a structural way how to interact with the world, right?

And the same should be applied for you, right?

The word is education.

You should have at least like the basics AI educations and the basic AI knowledge,

like something like the importance of data, about statistics, right?

What are the boundaries?

What are the possibilities of AI?

And what is the difference actually between machine learning

learning and deep learning, when should I use rule -based systems, when should I do all

these fancy words like what's an LLM, what are these vibe coding everybody's talking

about?

Agents, recs, right?

All these words that you have like knowledge, what are you talking about?

1And the goal for you should be taking confident decisions.

Phase 3: The Young Professional — Problem Solving with Deep Research

You need this knowledge for your next phase when you're becoming like a young professional,

for example and imagine maybe yeah that's that's you just just just started and you

may be working in the innovation part i know we have some people here from the innovation

hub for deutsche bahn lufthansa as well or you're in a consultant consultancy you're in a sales guys

and imagine that's you and yeah you have like this problem you want to identify

or solve problems because that's what we have to do we get paid to solve other people's problems

problems, right?

And since we're talking about AI, your idea is maybe to solve a problem

with an AI use case for maybe your company, but also maybe for a company you don't know.

Using Gemini Deep Research as a Second Brain

And to tackle this, I want to give a practical example with the Google Gemini deep research.

And maybe a question to you guys who have heard about the book from Daniel Kahneman,

on thinking fast and thinking slow raise your hands oh wow okay more or less everybody so for

those who haven't actually like he divided like the human brain in two different systems first

system yeah the thinking fast system system one it's like our intuition and the second system

thinking two system is the slow system where we really like thinking about um tasks like having

like complex problems and how to tackle these complex problems and if you draw like a scale

yeah then gemini deep research or also chat gpt whatever deep research functions are really like

on the left hand side second uh second brain yeah actually deep research is looking like this right

you have your classical prompt what is yeah actually we all know this from chat gpt and what

What is deep research doing then is really to ask you,

okay, is this really the problem you want to face, right?

And then you can adjust or start, like, the research.

It then will take, like, up to 15 minutes.

Maybe we can reset it again.

Up to 15 minutes, and it's really, like,

transparently showing you, like, the thinking process,

showing what kind of sources they're using,

how they're coming up to the solution,

Prompting Fundamentals and Tips

you're prompted and maybe that's basic knowledge but actually it's it's quite true because we use

this as well and for us it's really important to know that prompting here is really all you need

and I want to guide you through some tips to tackle this task I mentioned now first one is

to set up a role I think Stefano did it the same right so maybe like imagine like an AI expert in

in the public sector.

Next one is to set a clear goal.

What do you want the AI, the Gemini Deep Research,

to solve for you?

For example, here's I want to do a leadership workshop,

create comprehensive strategic analysis.

We want to have potential AI use cases to solve problems.

After that, you could provide it with a lot of data

and also the website of the company.

company, always GDPR confirmed, be aware what you can share your system and whatnot.

The more data, the better, and the more precise the data is, the better.

Stefano did this with his Firefly where he transcribed the meeting with Jacob, and actually

that's also like the best data you can get, right, to specify for a specific task.

mask.

Next one is really structured instruction.

This is what I would say like the core of

the prompt here for deep research analysis.

And there you can think of different types,

right?

For us, I took like these four different points.

Structuring Instructions and Output

Maybe like analyze like core processes.

Think of pain points.

Have a look of pain points.

Do your research on use cases, what

what it may be used for from other companies, right?

If you do like an innovation for Deutsche Bahn,

like how is Lufthansa innovating?

Maybe they have like something on their online videos

and so on and so forth, yeah?

Make an assessment of the AI maturity level

and try to be specified, right?

Maybe you can also go with,

please do the evaluation of these AI use cases

based on technical feasibility, strategic value,

you define KPIs and compare different use cases regarding these KPIs and then

the last thing is output format and a lot of people are forgetting this you

could also you we want to minimize this copy paste thing right and why don't

you use like an output format you give it already to Gemini maybe like a

markdown syntax that you can copy to your PowerPoint slides or whatever

Whatever slides, right?

Have like a use case title, always second level heading, you can have like please always

give me like a description, two sentences max, and so on and so forth, right?

So what you come up with then, maybe something like this, like an intelligent document processing,

and actually I did it for a client as well, and of course, yeah, you have to decide in

And in the end, is this really valuable or not?

Because hallucination is still a thing, right?

But in the end, once you put this information, you can paste it into your slides and really

have a starting point when you want to start a conversation with an unknown department,

unknown innovation center, to really do not start at the beginning anymore.

more so what i want you guys from this phase three is to never start from zero again okay so if you

come to a client you have a client call or so on and you you are not prepared um you didn't do your

homework okay do like this this deep research always have your human the loop your own intelligence

human intelligence in the brain yeah and take it into mind um because like some garbage will come

come out, some use cases are very valuable, you only have to do small adjustments and

put it together and then you may or less ace it.

So never start from zero again.

Phase 4: From Individual to Organization — Scaling AI Adoption

Last phase, and that's from the individual to the organisation, passing on your knowledge.

And we at Merantix Momentum developed this process, what is really applicable if you

You want to innovate AI -based companies, departments, or teams.

Education, Ideation, Validation, and Delivery

So we always start with education as well.

Not personal education, but educating the people who are relevant.

After education, you take on this knowledge and you ideate on AI solutions.

Because the client knows their problem best.

They're working there for years, whatever.

month so it's always good to co -work through these processes.

People can take

their AI knowledge and ideate on these problems.

Then is the validation part.

The validation part of AI use cases and then we prioritize these tasks and

afterwards we're going like into this all -known like POC MVP phase and into

to production.

A Holistic AI Canvas: Business, Data, Organization, Lifecycle

We at Merantix Momentum, we want to assess in a holistic way.

When you

think of ideating AI use cases, we think of these four different pillars, right?

We start

with the business value.

What is the business value?

It's basically business model canvas,

but with an AI switch on it.

Always think about data, data security, what data can we

use, what data do we have at hand, what is with IT security, do we have some ethical

problems on it.

Next thing, organisation.

How to bring these AI solutions really to the people.

Talking about expectation management, talking about adjustment to these solutions.

How to communicate AI.

Do I lose my job or will it accelerate my job?

And then put all this together in this AI lifecycle.

So we really have like a sustainable solution in the end.

You can download this AI canvas as well, it's like a 30 -page canvas, but the QR code is

also showing in the end.

Next one, prioritising.

Prioritization and Lighthouse Cases

So we always, after validation of the use cases, try to prioritise cases, right?

And then it's about you like how what kind of dimensions do you want to need?

What is applicable now for example technical feasibility was this business and strategic value?

And we always recommend to start with a lighthouse case.

We want to create this momentum

Okay in the company you want to get people to love AI you want to get people to you know

Start this AI transformation in a good way and not with a failure

So always try to find a quick win here

to start with

Afterwards you have your hypothesis.

Yeah with different possible AI cases and then you go through this normal process

We all know it POC MVP phase until production and then you iterate leave the try to improve

Conclusion and Key Takeaways

That's it for phase four.

So to round it up today's learning we started as a child right playing around

I want you guys to to get

When you go home like to really like just do it go on this website lovable and at end whatever build your stuff

without expectations

Educate yourself on basic AI so the goal should be doing confident decisions

Next one

Never start from zero again use this Gemini deep research

Approach or another approach

So I want you guys to never start from zero again because you don't have to I come to the question in a minute

Okay, and next one.

Yeah, AI transformation

It's taking a bit longer if you want to transform like whole companies in just a couple of weeks or months

You can start with the AI canvas, but we are also happy here today

To chat with you guys and we would love to guide you through this process

That's it for so far.

Thank you so much