Currently, there are two main trends in the AI market: AI as a product and AI as a feature.
AI as a product mainly comes from startups like OpenAI, Anthropic, Ideogram, or Runway. You create an account and usually pay a monthly subscription fee to use these tools. Often, there is a free trial access with usage restrictions, fewer features, and other limits.
AI as a feature, on the other hand, mainly comes from established companies like Google, Apple, or Adobe. They integrate it into their existing products. Currently, this is often included at no extra cost. After all, you are already paying for the use of these products, for example, with a subscription, through the purchase price of the device, or with your data.
Additionally, there is a third variant that is a kind of middle ground: AI as a service. As an analysis of the revenues of OpenAI and Anthropic has shown, this can already be important today. The keyword is: API, Application Programming Interface. This means that application developers can rely on the services of OpenAI and others.
I ultimately see today’s AI models most as a feature. I believe that the big tech companies are currently right: people want to use AI where it is useful.
Example: AI as a writing coach. Of course, I can copy an email into the ChatGPT app, ask the assistant for feedback, and copy the result back. However, it is much easier if that is a function of my mail application and is available there with one click.
It works similarly with Adobe’s new AI features: The image and video model Firefly is directly integrated into the corresponding programs. If I were to use a service like Runway for this, I would first have to upload my clip there, edit it, download it again, and add it to my project – cumbersome!
This does not mean, however, that it will always remain this way. The more powerful the AI models become, the more reasons there are to view and use them as standalone products.