How Generative AI Services Are Disrupting Platform Business Models (1 of 2) (Tech Strategy – Daily Article)

Posted on by

I did a podcast about how “Generative AI as a Service” is disrupting some particularly dominant platform business models. I mentioned YouTube, TikTok and OnlyFans. But we can also point to search engines, community sites and many others.

The question I raised was:

  • Are we finally seeing a business model that is more powerful than the digital platforms that have dominated for the past 15 years? Are big companies like YouTube, TikTok and others in big trouble?
  • Has a new super predator entered the business world?

I gave three examples. In these two articles, I’m going to go into them in more depth.

Example 1: Didi’s Platform vs. Robotaxis as an AI Service

Shortly after Didi went public, I wrote an article (When Will Didi Become Profitable? (Tech Strategy – Daily Article)) analyzing its business model and predicting that its business model would be disrupted in about 5-10 years. I argued that the emergence of autonomous driving would make much of its platform business model (which connects drivers and riders) obsolete. If you don’t need drivers, you don’t have a platform.

In that article, I wrote:

“This is the big technological disruption looming in Didi’s future. Once autonomous vehicles start doing robotaxis (note: taxis are Didi’s primary competitor), the cost structure drops dramatically. And it will keep dropping. That’s what hardware and software (and not human labor) does over time.

Imagine autonomous robotaxis giving rides 24/7. Constantly patrolling the streets of a city by the hundreds of thousands. They will be much cheaper. They will be safer. And the supply will be flexible, with more robotaxis coming off the assembly lines as needed. The system will maximize utilization per car and per fleet.

Recall, Didi’s ridesharing is a marketplace platform for services. It enables transactions between riders and drivers. But Didi’s bike-sharing business is a traditional pipeline service business based on deploying assets all over town. It’s more like a vending machine business than a digital platform. Autonomous vehicles will be the same. You won’t need the drivers anymore. And Didi won’t have a platform business model for this.”

So, what is Didi planning to do about this pending disruption.

First, it says this will only impact part of its business. There will still need to be human drivers as well. Ok. That is probably true but not a great future to bet on.

Second…

“Didi says it is developing Level 4 autonomous driving with its partners. And it aims to be the operating system.”

So Didi wants to leverage its current position into being the operating system for autonomous driving. But they don’t want to create cars. They want to use their current demand-size power and partner with car manufacturers. And Didi (like Uber) certainly has customers and demand today.

It’s a standard marketplace platform. They have lots of demand and lots of data and engagement. But let’s not kid ourselves. Robotaxis are an entirely new business in its earliest days. Didi has no real advantages over the many other companies going after autonomous vehicles. And while Didi says it is going for a hybrid supply (drivers plus autonomous), this game could end up being mostly autonomous.

So, how is this going to play out? Can a platform business like Didi end up controlling the demand, data, and operating system of an AI service? And then the car manufacturers are just going to be suppliers? That would be like the Android model, with lots of companies making handsets but Google controlling the operating system.

Or maybe Didi is just going to offer its own AI Service business of robotaxis? It will not be an operating system for everyone. It will just have an AI service plus a platform itself. Which would look like this:

You can see the standard platform on the right, which serves riders. But you can also see a linear, pipeline business model also serving them. And its key activities are AI transportation services.

But which will riders prefer? That is the key question for this article.

  • When is an AI-powered service superior to a platform business model?

I think the future of this business is an AI-powered service. Didi, Uber, Lyft and Grab are going to all transition to AI-services if they can. But this is going to be a long and expensive fight. And it’s pretty clear they are going to need to partner to be viable competitors.

***

Ok. That is example #1. And it’s a good one to start with because it’s a pretty simple AI-powered service. When we look at “Generative AI as a service” taking on other platforms, we see a more complicated version of the same situation. Generating content is more complicated than just doing a taxi service. And it’s more powerful. But it’s the same AI-as-a-Service vs. Platform Business Model question.

Example 2: Google’s Search Platform vs. Neeva’s “Generated AI as a Service”

One of the first questions raised when ChatGPT appeared was whether it would replace Google Search. With it, you can ask any question you want and get an answer – and not just a list of links. I have described it more as an alternative user interface between humans and the information on the internet. So, it could be considered similar to newsfeeds, which is another human-machine interface. Others have started to describe it as “mansplaining as a service”, which is pretty funny.

Google Search is a bit complicated as a platform business model. I don’t think I’ve ever really detailed it in these articles and podcasts. I describe it as a learning platform (one of my 5 types). And I use the below graphic.

Let me walk through this as it is a bit complicated.

I have detailed 5 different types of platforms based on the types of interactions they enable. Marketplace platforms enable transactions. Payment platforms enable payments. And learning platforms enable interactions that make the service smarter and better. The key activity that drives the learning is the interactions. I suppose you could call them Q&A platforms as it’s about enabling question and answer interactions. Examples of these are search engines like Baidu and Q&A sites like Quora.

For a search engine it is enabling interactions between people doing searches and webpages full of information that are created and updated by people. Searchers and webpages are the two user groups. This is similar to how marketplaces are between consumers and merchants. The second party (webpages) is created and maintained by people. Like merchant stores. Although this is getting more complicated because the webpages are increasingly being created by both human and digital agents.

But the platform’s role is to aggregate demand and supply and then do the matching that enables the interactions. Note: We also see a standardization network effect on the supply-side because people create their webpages to make them easily searchable by Google’s SEO.

OK, so how does ChatGPT compare to Google Search?

First, you have to separate GPT3, which is the technology, from ChatGPT, which is a public service built on GPT3. GPT3 (soon GPT5) is an innovation platform. It is something that app developers build upon, just like Microsoft Windows and the App Store. Right now, thousands of apps are being built by tapping into the GPT APIs and using their functionality. And ChatGPT is just one of these applications.

Second, GPT as a large language model gets better and better based on human usage. So, there is a bit of a network effect there. The key technology behind LLM is “reinforcement learning with human feedback”. So early movers do have an advantage depending on how complicated their service is. We should see other LLMs, but they are expensive so there may not be many. We may end up seeing lots of smaller language models (probably built on top of an LLM).

Based on this, the business models looks like “Generative AI as a Service”.

  • It’s a linear business model that generates personalized content on-demand.
  • You ask a question or give it a prompt (write me a poem) and it generates, in real-time, an answer that is, in theory, accurate and personalized to you.
  • You can offer repeated prompts to refine the content. And it will remember your past interests and requests and increasingly personalize to you. It starts to look like personalized service for information and other content creation.

***

So, the question is really:

  • How do Generative AI service businesses (like ChatGPT) built on LLM innovation platforms (like GPT) compare with Google’s learning platform?

That’s kind of wordy. But that is really the key question. Read it again.

To me, it looks like the future for Google Search is going to be similar to Neeva. This is a new company that basically combines a ChatGPT type service with a search engine. You type in your search or question, and it generates an answer. And below that answer, it provides search results. Here’s a typical result from Neeva.

You can see it answering my question directly. And it cites the sources it used to generate the answer, which is helpful.

But it also has the results similar to a search engine. You can see the videos, personal and images tabs on top. And it gives search links below.

So, it’s a platform plus an AI-powered service (sort of like Didi’s dream). It looks like this.

So, Which Will Be the Stronger Business?

We know the answer for robotaxis. It’s just a better service.

But when using Neeva will people mostly just look at the AI generated result at the top? Will they still go to the links below? Note: Google only makes money when you click on the links in search results. So, this service is a direct hit to their revenue.

My prediction is “AI as a service” is going to become the dominant user interface for most types of active inquiries. We will see three types of interfaces between humans and the information of the internet.

  • Search engine – which is an active inquiry. This is static and one-directional. You get links.
  • Newsfeed – which is a passive consumption of streaming content.
  • Generative AI services – which are active inquiries. This is dynamic and becoming two-directional.

I think all three will exist. But I think Newsfeed and Generative AI are most of the future. That’s my guess.

Which brings me to Example #3, which is really the big one. What happens to YouTube, TikTok, OnlyFans and others?.

That’s what I’ll detail in Part 2.

Cheers, Jeff

Also, here are some other search engine graphics.

————

Related articles:

From the Concept Library, concepts for this article are:

  • Generative AI
  • Learning Platform
  • AI-Powered Service Business Model
  • Pipeline Plus Platform Business Model

From the Company Library, companies for this article are:

  • OpenAI / ChatGPT
  • Didi
  • Neeva
  • Google Search

Photo by Mitchell Luo on Unsplash

———–

I write, speak and consult about how to win (and not lose) in digital strategy and transformation.

I am the founder of TechMoat Consulting, a boutique consulting firm that helps retailers, brands, and technology companies exploit digital change to grow faster, innovate better and build digital moats. Get in touch here.

My book series Moats and Marathons is one-of-a-kind framework for building and measuring competitive advantages in digital businesses.

This content (articles, podcasts, website info) is not investment, legal or tax advice. The information and opinions from me and any guests may be incorrect. The numbers and information may be wrong. The views expressed may no longer be relevant or accurate. This is not investment advice. Investing is risky. Do your own research.

Leave a Reply