Tobias Zwingmann

RAPYD.AI

Independent AI consulting to create human-centric AI solutions that innovate and grow your business. Fast.

Managing Partner

O'Reilly Media

Author & Online Trainer

Springboard

- Helping Springboard as a subject matter expert in AI, especially for data analytics
- Mentoring students who want to switch careers to data science / data analytics 
- Top Mentor Award in 2021 (Springboard Inspire)

AI SME & Data Science Mentor

Toptal

Independent Consultant

Data Science Dojo

ML/AI Instructor

Deutsche Messe AG

Systematic retrieval, analysis and utilization of (unstructured) data from diverse sources to build new products and processes.

- People movement analytics 
- Computer vision with focus on text retrieval

Senior Data Scientist

Building & maintaining predictive models. Generating and communicating insights about customer trends and behavior to product owners and senior management.

- Churn Prediction
- Next Best Offer Recommendations 
- Customer Segmentation
- Visual Analytics

Data Scientist

Leading various international trade show projects in e.g. China, India, Indonesia, Mexico, Myanmar, South Africa, South Korea, Turkey.

Project Director

Deutsche Messe Dubai Branch

Management Trainee

Author of AI-Powered BI (O'REILLY) | AI Coach & Mentor | I help organizations leverage AI and use data more intelligently to improve business performance. Sharing AI use cases & learnings.

As an ex-Senior Data Scientist, I've seen the power of AI to revolutionize businesses. But I've also seen the pitfalls of a tech-first approach, especially for nontechnical B2B companies:

- Building customer churn prediction models with 95% accuracy? Check. 
- Implementing a next best offer system with 15% uplift? Been there. 
- Innovating a data-driven B2B product that delivers 6-figure revenues right from the start? Done that. 

But over the last 10 years of my Data/AI/ML journey I've realized that these results ultimately don’t matter. 

What matters is how you can leverage technology not for a single use case, but for an entire organization. 

The key to this transformation: People. 

I’ve worked with businesses around the globe, from B2B SaaS startups, over brick-and-mortar SMB's to leading financial institutions. 

The common theme?

Empowering people, not replacing them. 

That’s how I typically help my clients achieve positive ROI from their AI investments within the first 3 months working with me:

- achieving higher productivity, 
- innovating better products,
- growing their business.

Today, I'm now on a mission to help B2B leaders use these principles of Augmented AI to thrive in an AI-driven world, and future-proof their business - without hiring more tech resources or breaking the bank. 

A bunch of my insights is shared in my books "AI-Powered Business Intelligence" (O'Reilly 2022) and "Augmented Analytics" (Co-author, O‘Reilly 2024) as well as in my weekly newsletter "The Augmented Advantage" that is read by 4,500+ business leaders from brands like Amazon, Mercedes-Benz, Gucci, and Santander. 

So if you're a B2B business leader who's tired of seeing AI through the tech lens only, but wants to thrive in an AI-driven world, let's talk. 

I'll show you how Augmented AI can be your competitive edge.

Message me or connect!

We help you build amazing AI products faster.

Meta (Facebook) just spent over $100,000,000 to build Llama 3 - the world's best open source AI model - so you can use it for free. Here's why they did it and how you can benefit:

Meta is the only Big Tech company committed to developing AI, particularly large language models, with an open-source approach. Why? My take: it's less about altruism and more about strategy. By making LLMs essentially free, they're aiming to crush all commercial competition.

Llama 3, Meta's latest model, was trained on 24,000 (!) GPUs (each costing around $40k). You can do the math.

Two versions are immediately available:

- Small (GPT-3.5 class, 8B)
- Medium (below GPT-4 class, 70B)

The large model (GPT-4 class) is still in training and will be released later this year.

I've played with the medium-sized model and I'm really impressed with the results. Meta has built a great open source model with really good reasoning and logic capabilities.

There are 3 ways you can use Llama 3 for your business:

1/ Llama 3 as a Service
Use Llama 3 from any cloud provider as a service. You pay by use, but the price is typically much cheaper than proprietary models like GPT-4 or Claude.
→ Use Llama 3 on Azure AI catalog: https://lnkd.in/eptszsUD

2/ Self-Hosting
If you have GPU infrastructure (on-premises or cloud), you can run Llama 3 internally at your desired scale.
→ Deploy Llama 3 on Amazon SageMaker: https://lnkd.in/ext8cmBH

3/ Desktop (Offline): 
Tools like Ollama allow you to run the small model offline on consumer hardware like current MacBooks.
→ Tutorial for Mac: https://lnkd.in/eKnPbHne

Bottom Line: There’s really no point any more in trying to train your own large language model from scratch unless you’re in Big Tech or AI research. 

Just take the models that are out there, fine-tune on your data (if necessary) and start building!

Will you use these models to augment workflows, increase productivity or simply be more creative in your business? 

The choice is yours - but the opportunity is too big to ignore.

Have you tried Llama 3 yet?

AI Augmentation Advisor | Author | Speaker - Helping forward-looking B2Bs innovate and grow with AI (no big tech team needed). Featured on CNBC. New 📕 "Augmented Analytics" (O’Reilly) incoming

I've seen it time and again - businesses that try to do too much, too quickly with AI end up crashing and burning. They try use cases like:

- AI for automated data analysis
- AI for automated email response
- AI for automated lead flow optimization 

To be crystal clear:

Starting your AI journey with any like these is a horrible idea because they require high automation and high integration. 

That’s the most difficult combination to crack and even highly mature AI companies struggle with this. 

But where SHOULD you start?

Take a look at the following matrix. 

I call it the Integration-Automation AI framework. 

No matter the size or industry, I always use this as part of my onboarding process with B2B companies that are looking for my advice on their AI use case. 

The framework helps us figure out where their current use case sits and where it should be, given the AI maturity of the business.  

Want a detailed breakdown?

Check the comments👇

Your IT department isn't the key to AI success

Your IT department is already overwhelmed. Not only do they rarely have a clue about AI (no offense), but more importantly, they are already in firefighting mode on 1,000 other issues. Adding AI to their overflowing plate is a recipe for disaster, just like outsourcing "data" to IT 10 years ago.

We've seen this mistake before with Big Data.

Companies hired data scientists, stuck them in IT, and wondered why the "business value" never materialized.

And now, those same companies are struggling to keep up in a world with AI.

Don't let history repeat itself.

When it comes to AI, look beyond IT.

To find out where to look, check my upcoming newsletter tomorrow.

Sign up here before it goes out so you can innovate and grow your business with AI - without being limited by IT: https://lnkd.in/egURH-BZ 

AI success isn't about having the fanciest tech 
AI success isn't about having the largest team 
AI success isn't about having a killer use case
AI success isn't about having a PhD

AI success is all about having the discipline to stick to a strategic plan and the agility to adapt as you go. The high-level roadmap for it is this:

1. Understand the AI basics
2. Identify business problems
3. Map AI opportunities
4. Analyze use cases
5. Prioritize and prototype
6. Iterate and improve

Forget the hype. 

These six steps are really your key to AI success. They sound so simple, and yet they are so hard.

The real challenge isn't in knowing the steps. 

It's in actually executing them. 

It's in staying the course when your CEO picks up the latest AI buzz from a conference. 

It's in holding steady when your competitor launches their "revolutionary" GPT knockoff.

It’s in navigating a path that’s littered with distractions and dead ends. 

Don’t get sidetracked. The time you lose is time your competitors win.

So claim this roadmap, make it yours and stick to the plan.

That's how you grow your business with AI and claim your very own Augmented Advantage.

How far are you on your AI journey?

I’m speaking at HANNOVER MESSE next week. It’s the world’s leading industry show with 100,000+ visitors from around the globe, i.e. "the place to be for the tech to come". I’ll share some of my observations and lessons on how to (and not how to) innovate and grow with AI, especially if you’re a traditional company without a large tech team. 

Details:
- Date: 23.04.2024
- Time: 09:45 - 10:15 (CET)
- Where: Industrial Transformation Stage
- Location: Hall 3, Stand A76

If you’re there, let me know. I’ll probably be around the whole day!

Data analytics is one of the most underrated ways to augment your own capabilities with AI. Why?

AI in data analytics can help you to:

- Get answers from your data in natural language
- spot hidden patterns and relationships
- highlight trends and anomalies 

so you can 

- get faster insights
- discover novel ideas
- reduce chances to 'overlook' something

Sounds complicated?

It isn’t!

In fact, everyone with a Microsoft Office subscription can use AI in PowerBI without any coding, extra costs or administrative overhead. 

For example, the screenshot below shows Power BI's built-in AI anomaly detection, which not only flags unusual patterns, but also runs an analysis across all fields in your data to find possible explanations. 

If you want to learn how to do this + 3 more use cases for AI in diagnostic analytics, sign up for my upcoming online live training with O'Reilly on April 11, 9am PDT: https://lnkd.in/ew9-U-SX

There will be a recording, but it’s only available if you sign up BEFORE the event. (I can’t send the recording afterwards.)

Just a few days left to sign up! ⏰ 

PS: Participation requires an O'Reilly subscription. If you need one, comment COUPON below and I'll send you a free 30-day access pass. 

Image credit: Microsoft

"Amazon's Just Walk Out stores relied on '1,000 people in India,' not AI."

This news went viral last week and I think it was often taken out of context. 

It left the impression that AI is fake and AI doesn't work. 

But what most people don't understand is that there are two types of AI:

1. AI cases that are largely solved, like using computer vision to extract information from documents.

2. AI cases that are extremely challenging, like autonomous driving or recognizing items grabbed from a shelf, which tech giants are spending billions to figure out.

As a "normal" business, don't waste time on use cases that need category 2 AI. Leave that to big tech. Instead, focus on category 1.

Identify AI services that work well out of the box, and consider how you can adapt them for your business. 

You may need to give them your own spin, like fine-tuning the models. This is how you can reap the benefits of the billion-dollar investments made by tech giants.

Because if you don't, your competitors will.

Watch Philipp Neuberger and me discuss this in more detail below!

What's your take on outsourcing 'AI' to people?

---
PS: If you found this discussion valuable, please share or repost! ♻️

Guten Morgen HANNOVER MESSE! 

Danke Tobias Zwingmann für die super spannenden Einblicke in die Herangehensweise an KI-Projekte. 
„Keep humans in the lead, not in the loop!“

Mein Kollege Rodrigo Groener und ich sind heute (Dienstag) hier unterwegs. Falls jemand aus dem Netzwerk auch hier ist, schreibt mir doch gerne kurz eine PN 😊

#KI #hannovermesse

AI isn't about make OR buy. It's make AND buy. 

You might think that AI is just another software project. But that's not true.

AI is another software project, plus… well… AI!

In traditional software development, the roles were pretty clear:

Business says what they want.

IT delivers it. (Hopefully)

There might some cross-collaboration here and there, but essentially there was a clear boundary.

Essentially, you didn’t care HOW the IT guys were doing the thing.

With AI, however, things have changed.

For the average, traditional B2B company, training / building their own custom AI models from scratch is completely out of scope because of a lack of:

- data
- talent
- infrastructure
- organizational readiness

Given these limitations, the unit economics for most "build your own AI" use cases just don’t make any sense.

So instead, the default mindset should turn from "Which AI do we need to build?" to "Which AI is out there we can leverage to solve the problems we have?"

From there, you can mix and match on different scales like:

- Ownership
- Flexibility
 - Control
 - Speed
 - Talent

Depending on your use case requirements, criticality and internal capabilities.

This could mean:

- Just buying a ChatGPT Teams license for a non-critical workflow
- Leveraging an AI-as-a-service platform for a small quick-win
- Using open-source models internally if you need more scale

The point is, you have options. 

But 99% of cases your internal IT will come back to you - the business - to get the answers they need to navigate this decision spectrum.

So it’s up to you - the business - to not only understand the problem space but also the solution space to a certain degree.

That's why I do an "AI Basics" crash course Day One when I work with new clients.

Because when you know what you want, you can work with IT much more effectively.

And get positive ROI on your AI projects in 3 months or less.

I left some more tips in the comments!

Quick reminder:

When you want to build an AI strategy and you hire an IT consulting firm to help you, what you're getting is another IT project disguised as an AI strategy.

Happy weekend!

What impact will AI have on the workplace?

With modern AI, we don't need more humans in the loop. 

We need more humans in the lead.

The problem with most AI solutions is that they treat humans as an afterthought. 

"Just another cog in the machine"

But:

People are your greatest asset. 

They shouldn’t be treated as a system component to 'oversee' the AI and correct its mistakes.

So consider this mindset shift:

It's not about plugging AI into your existing processes and hoping for the best. 

It's about putting your people at the center of your strategy, operations, and innovation. 

And THEN fueling people's creativity and skills with AI. 

This shift from "human in the loop" to "human in the lead" is radical, and it spans much further than just AI. 

But when you do it this way, you truly have a chance to transform your business (instead of making it just another empty catchphrase on your strategy deck).

AI is a tool to empower your people, not the other way around.

Are you ready to put people center stage?

——

PS: If you want to learn more about human-led AI solutions so you can grow and innovate your business, join my newsletter: https://lnkd.in/eFzzQrMJ

AI automation aims to replace people. 

AI augmentation aims to empower them. 

Which do you think your team needs more?

AI is no longer optional, but the way you implement it makes all the difference. 

Automation seems like the easy choice. Let the machines do the work, right? But at what cost?

Automation may boost short-term efficiency here and there (if you’re lucky). But in the long run you risk losing your creativity, intuition and human touch that make your business unique. 

Augmentation, on the other hand, is about enhancing your team’s capabilities. It’s about giving them the tools and insights they need to do their best work. 

With Augmented AI, your team can work smarter, faster, and more efficiently - without sacrificing the human element.

That’s the future I want to build with Al. 

Because no matter how you look at it:

In the long run, if you don't learn to leverage AI, your business has no chance to survive. 

But let’s use it the right way right from the start.

Volle Hütte beim genialen Vortrag von Tobias Zwingmann auf der Hannover Messe 2024 mit sehr passenden Memes.

Die Zusammenfassung:

How to fail with AI

Phase 1: Hellhörig geworden: Chefetage hört auf einer Messe von KI.
Phase 2: Angezündet für das KI Thema: Chefetage beschließt KI muss in der Firma eingeführt werden.
Phase 3: Personal wie Data Scientist, oder Beraterteam wird gesucht und eingestellt. Dem Thema wird viel Zeit und Geld gegeben.

Ergebnis: Zeit vergeht. Am Ende kein erfolgreiches KI Projekt. Sehr viel Zeit und Geld investiert, ohne Erfolgsergebnis.

How to succeed with AI

Erfolgreiche KI Projekte beruhen auf der Disziplin, sich an einen strategischen Plan zu halten und der Agilität, sich im Laufe der Zeit anzupassen. Der übergeordnete Fahrplan dafür lautet:

1. Die KI-Grundlagen verstehen
2. Geschäftsprobleme identifizieren im Brainstorming
3. KI-Chancen abbilden
4. Anwendungsfälle analysieren
5. Priorisieren und Prototypen erstellen
6. Iterieren und verbessern

Starke Empfehlung: Prototyping für KI Projekte

In kleinen iterativen Schritten vorgehen. 
So klein wie möglich starten, z.B. mit einer banalen KI Chatbot Lösung beispielsweise ein FAQ Chatbot und mit schnellen Ergebnissen bei den Prototypen Punkten.

Auch Tobias hat man angemerkt, dass er sehr viel Zeit mit ChatGPT verbracht hat. Das Wort "Gamechanger" viel. 😉


Tobias #HM24 #Linvelo

Taplio

Tobias Zwingmann's Linkedin Analytics