Your Guide to the AI 2.0 Gold Rush

Good morning. Fall is here, and so is the year-end rush to deliver our overambitious roadmaps, which didn’t take into account that people will be scrambling to prepare for holiday trips, attend all the holiday happy hours, and get the scarce supply of the popular Eggnog from Costco before it is sold out.

What’s left is to wish for Santa to make the miracle of delivering all the features by the end of the year come true.

Also, with fall starting, the most recent S24 Y Combinator batch shows off what they are building. It is a good opportunity to see what startups are doing, and one thing is clear: they all use AI. Some show an evolving set of use cases that point to new opportunities across all industries.

Techcrunch provides a nice summary of Demo Days One and Two. These are very early stages, but they show where AI as a foundation is going.

It’s not just improving models; AI is becoming more than a text companion. It’s the artist, the doctor, and the entrepreneur. As many now call it AI 2.0.

In today’s post, I will be looking at a few industries and what some more mature companies are doing, how they leverage AI, and what benefits it brings.

Today’s goal is to explain AI 2.0 and inspire you to discover new opportunities.

See it as a preparation for the AI 2.0 gold rush.

What is AI2.0?

So, is ChatGPT-4 an example of AI 2.0, and if not, what is?

ChatGPT is part of it. It is the evolving generative AI that powers AI 2.0. As we know, it creates human-like text, can answer questions, write stories, or generate code.

It processes information and generates content.

AI 2.0 goes beyond that. It covers capabilities that expand outside of large language models and find much more sophisticated use in the real world and real-world interaction,

AI is not one technology. It is a collection of tech that drives forward in parallel.

A bit of Theorie around Evolving AI Techniques

• Reinforcement Learning is a type of machine learning in which the models learn via trial and error. The AI receives rewards and penalties based on its decisions.

  • Use case: Optimizing repeatable processes like robotics in manufacturing

• Transformer Networks are deep learning models designed to handle sequential data. Like text inputs. They make sense of text by understanding the relationship between words.

  • Use case: Large Language Models (LLMs) are based on transformer networks.

• Federated Learning helps train machine learning models across multiple devices and connects them through the cloud. No raw data is shared, which improves privacy. It is a decentralized data model.

  • Use case: Locally trained AI models based on your data. Like Google’s Gboard. It is personalized but personal data stays on the device.

• Causal Inference determines cause-and-effect relationship from data. It wants to find what causes specific outcomes.

  • Use case: AI can run “what if” scenarios and is helpful in healthcare treatment evaluation, such as determining what drug causes improvements in patients.

• Synthetic Data is like a data supplier for models. Once AI has been trained on all available data, it starts to create its own. We talked about the risks of this in a recent post: https://www.techtalesandtactics.com/p/ais-diminishing-returns-innovation-needs-reboot

  • Use case: Autonomous vehicles and computer vision learning based on AI-generated images, for example.

AI Startups We Should Pay Attention To This Fall

Let’s look at a few real-world examples to get inspired for new opportunities.

1. Healthcare is making fast gains

I am no expert in this, so I had to read up and learn to give you some value here. But that’s the whole point. We learn something together.

Drug discovery is a costly investment, and it can take 10+ years to bring a single drug to market. In this “drug-to-market” journey, AI can help in several ways.

Early stage discovery - AI can screen millions of compounds in a fraction of time. AI can also help prevent dead-ends. Speeding this process up can cut costs by 30-50%. Companies that work on this:

• Insitro - started in 2018, total funding of $643 million

• Atomwise - started in 2012, total funding of $174 million

Clinical trials - AI can help identify candidates better as well as predicting side-effects. Clinical trials are the most expensive part of the “drug-to-market” cycle. And any optimization can save big. Companies driving innovation here:

• Owkin - started in 2016, total funding of $321 million

• Unlearn.ai - started in 2017, total funding of $135 million

2. AI Sales Reps, Customer Outreach, and Lead Generation

If you work in sales or know the work of generating qualified leads, you will love these startups. They free up time for your sales staff and promise ROI boosts by up to 10X.

• Kalendar.AI - B2B lead generation and cold outreach. It can personalize mass emails, reach prospects and schedule meetings with pre-qualified leads

• Apollo.io - Automates multi-channel outreach - again, with personalized messages across LinkedIn AND based on the prospects behavior.

• Seamless.AI - Builds you a massive list of decision makers in the companies you are trying to reach. Integrated with LinkedIn and Salesforce, it has all the info it needs to do that.

3. AI is Glowing up the Entertainment and Gaming Industry

Every new technological breakthrough enables something new in the entertainment and gaming industry. When the iPhone gained traction, the first thing we saw were simple mobile games. And we saw them evolve into a billion-dollar industry. In 2024, the Mobile Games market is projected to generate $98.74 billion worldwide. Started less than 20 years ago, it already makes up about 50% of the total games market.

Then, we rushed into VR games, which have yet to cross the chasm to the mass market.

AI can bring the next big change for both gamers and developers. It’s hard to foresee, but the impact could be monumental. It could reach the point where simple games are entirely created by AI. And, as we see in other industries, they can also be branded and distributed entirely by AI. One-person gaming studios will be a reality.

For the consumer, that means a lot more options. But also a lot more noise.

We are already seeing big game studios use AI to create deeper interactions with non-playable characters (NPCs) within a game and create objects and buildings. I was very close to jumping into Warzone to see it… “for research.”

Interesting companies in gaming and entertainment:

• Series - A whole new game editor and authoring engine is emerging around these movements. It sounds like a platform built around the game development cycle and adding options to speed up development by using AI to build stories, interactions, objects, and items a player can use.

• Runway - Generating production-ready special effects and renderings in no time. They are already in talks with Lionsgate about using it for movie and TV production. I linked to their site with use cases and stunning examples.

  

• Stability.ai - seems like a direct competitor to Runway. What stood out to me with them was their Board of Directors. How can they fail with James Cameron and Sean Parker in their BoD?

4. Logistics: Mobile robots are taking over our luggage

Upfront, this is not in the same vein as the genAI, LLM advancements we see above. It is a more difficult form of AI. Traditionally, it combines Lidar, GPS, cameras, and SLAM algorithms.

But if we want to combine it with the power of large language models (software), we need to bring the world of software and physics together with mechanical engineering. A whole other level.

AI can also be used to enable swarm intelligence. Imagine hundreds of robots servicing a warehouse. Connecting them to each other in the cloud lets them all organize their work into efficient sorting, retrieval, and delivery and coordinate their movement overall.

•  Unbox Robotics - started in 2019, total funding $12.3 million

• evoBot - I couldn’t find much about the commercialization of this project. They seem to be in the test stages with the Munich Airport now.

The evoBot is the development of the Fraunhofer Institute and is now in a collab with the Airport Munich.

The evoBot's design is smart. It can accomplish a variety of tasks, such as loading shelves and doing service jobs like watering plants, in addition to self-pickup and transportation. It extends into all kinds of applications for material flow and logistics. For them adding the components of swarm intelligence, a software management platform and adaptive learning can revolutionize logistics in many places.

• Nvidia is betting on Robotics too - This is an area Nvidia CEO Jensen Huang doubles down on. At Computex in Taiwan he recently said there would be two “high volume” future robotics products. Self driving cars and the second is likely humanoid robots. Nvidia’s robotics stack is gigantic and a huge bet for them. Maybe something for another episode of this newsletter.

💥 Relevant breaking news during edit

This week, thousands of dockworkers along the East Coast and Gulf Coast went on strike. One of their demands was that automation on docks be completely banned.

They are asking for protection from automation, which has already taken over at some major ports. For some reason, this brings recent AI development into the discussion. But this is not new. Don’t get it mistaken. It is technology from more than 20 years ago—container trucks driving around a port following guided tracks. It’s a streamlined process with a limited set of decisions.

Conclusion and New Opportunities with AI 2.0

This weeks post was a journey. The goal was to look into a few interesting applications and companies that are building on top of the current AI stack.

While researching and writing, I made an astonishing discovery. Some industries are evolving so rapidly that their operations have already changed to some extent and will be completely different in the mid-term.

There is excitement but also a quick reality check for everyone.

AI 2.0 is going beyond data crunching and rolling out its upgrades to the world. We saw many examples that completely change industries already.

In the new AI gold rush, we trade shovels for algorithms and Nvidia GPUs.

I'm curious: Are you already involved in this in some form? I would love to hear your stories. And if not, I hope this AI pulse check got you thinking about how you can stake your claim in the future of AI.

Have a great rest of the week,