The battle for generative and conversational AI will be won on the API front

Representation of the "evolution of man' except with robots, representing the evolution from an API to an upright robot walking.

A strange title? Perhaps, but ChatGPT’s creators have demonstrated how well they understand what is at stake in the exploding market for artificial intelligence models. Their affordable, monetizable API announced this spring could be a game-changer.

But even though ChatGPT is capturing 80% of the tech media’s attention at the moment, it is important to understand that OpenAI is not alone in this field: other web giants are mounting an imminent counterattack.

AI war of the tech giants

OpenAI has certainly won one battle: the company’s perfect timing has transformed a lab innovation into something for the use of the general public. One innovation, one use, one audience, and word of mouth did the rest.

But neither Baidu nor Google will accept being left behind.

Alphabet, Google’s parent company, is opening access to “Bard,” which capitalizes on the advances initiated by GoogleAI and Deepmind and uses the latest conversational intelligence (LaMDA).

Oriented towards web search, Google’s DNA, it is positioned as a direct competitor to ChatGPT. Let’s hope it’s a bit stronger in math than ChatGPT!

On the Asian side, Baidu has published its own tool in direct competition with Dall-E’s AI-generated digital images, ERNIE-ViLG, and AliBaba isn’t far behind with its own ChatGPT-style chatbot.

Let’s be clear, beyond the training samples and the constraints of avoiding cognitive biases; the battle will be won on the ability to integrate these conversational and generative agents into business processes or customer journeys.

Microsoft understands this, having injected several billion dollars into OpenAI. Its little brother of the now defunct Cortana, based on ChatGPT, will enable professionals to obtain an automatic meeting report at the end of a Teams meeting.

Opening up AI capabilities via APIs

Beyond the ability to make this AI “useful” by exposing it as an API and introducing it into existing processes, it is above all the ability to monetize this AI directly that is pushing editors to offer it as an API.

From there, developers can build their own AI-powered applications or products; or use them to enhance existing products or services. Many are already doing just that, exploring these capabilities to get ahead of the AI wave:

  • Natural Language Processing (NLP) APIs: Many companies are using NLP APIs to analyze customer feedback and reviews, automate customer service interactions, and enhance chatbots and virtual assistants.
  • Image Recognition APIs: Image recognition APIs are used by e-commerce companies to identify and recommend products based on user-generated content, such as images and videos. Social media platforms can use AI-powered image recognition APIs to improve their photo tagging feature.
  • Speech Recognition and Text-to-Speech APIs: Companies are using these APIs to develop voice-enabled interfaces for their products and services, such as voice-activated assistants and virtual agents.
  • Predictive Analytics APIs: businesses leverage predictive analytics APIs to analyze large volumes of data and make predictions about future trends and customer behavior. This information can be used to improve marketing campaigns, optimize supply chains, and make better business decisions.
  • Sentiment Analysis APIs: these help companies understand customer opinions and emotions towards their products and services, allowing them to tailor their marketing and customer service efforts accordingly.

These AI APIs – and other uses cases yet to be discovered – offer businesses a powerful way to leverage AI technologies without having to invest in costly in-house development and expertise.

Monetization: the final frontier

Providing access to AI models and algorithms via APIs allows developers to build their own AI-powered applications or products, certainly, but the providers of those APIs have an opportunity to market and monetize these APIs. This is where the real value will lie.

Open innovation boosted by AI will involve exposing an API, yes, but not a technical API: a real digital product that can be consumed by all the developers in the world to boost their digital services. And it is the marketing capacity to distribute this digital product that will make the difference.

Offering AI APIs on marketplaces makes it easier for developers and businesses to find and integrate the AI-powered tools they need – expanding API adoption. And monetization can look like charging a fee for API usage, based on factors such as the number of API calls or the amount of data processed, although other indirect monetization models exist as well.

In March, OpenAI fired the next volley on the monetization front, releasing ChatGPT and Whisper APIs that make the technologies easier to integrate and affordable enough to support new use cases. OpenAI says it managed to achieve a 90% cost reduction for ChatGPT since December through a series of system-wide optimizations, which allows it to pass those savings on to API users.

This expanded access will no doubt also lead to more widespread adoption; OpenAI has taken its valuable NLP model and, capitalizing on its newfound popularity, packaged it into an API product that will generate new revenue and drive innovation.

That is the driving objective behind Axway’s own Amplify Enterprise Marketplace: a customizable API marketplace that provides the tools to make the difference between a simple digital asset (APIs) and a true digital product. As we’ve learned when it comes to APIs, platform companies may already have many essential building blocks in production, but the real value of digital assets lies in their consumption.

Already, ecommerce platform Shopify has announced a ChatGPT API-powered search feature for its Shop app, and a grocery delivery and social media platform have followed suit with their own implementations of the API.

Some closing thoughts on the AI revolution

There is a parallel between these AI-driven advances and the automotive revolution, which was driven by the internal combustion engine. In both cases, they started as complex processes reserved to technicians (mechanics, data scientists) which were then democratized, merging into tools that are easily manipulated by the average human.

Today, most people drive cars without knowing anything about a stoichiometric mixture or an ignition advance… in the same way, thanks to APIs for AI, we can use ChatGPT without understanding how a Turing machine works.

What can be frightening is the empowerment of these innovations: a self-driving car might be cool and even promise tremendous benefits for safety and efficiency, but autonomous vehicles have yet to reach mass adoption due to persistent liability, regulatory, and security concerns. And an autonomous AI, which can guide knowledge, quickly reminds us of HAL, 1984, or Terminator.

Some are already sounding the alarm bell on untethered use of AI. OpenAI, for its part, has not remained insensitive to security and privacy concerns, announcing last month the ability for users to turn off chat history. It remains to be seen how we will face these ethical dilemmas.

What is clear, however, is that those who can see the difference between AI as a technical capability and AI as a digital product – which requires its own marketing just like any other product – are the ones who will win the battle of AI monetization.

Read my follow-up article in this series on the AI revolution: Could ChatGPT Become My Advisor? [AI for Banking as a Service]

What is your API product intelligence? Discover our checklist for 10 KPIs to support your enterprise API strategy.

Key Takeaways

  • The battle for conversational and generative AI will be won based on the ability to integrate these agents into business processes or customer journeys.
  • OpenAI's affordable and monetizable APIs could be a game-changer in the market for artificial intelligence models.
  • Competitors, including Baidu and Google, are mounting an imminent counterattack in the field of conversational and generative AI.
  • Because the real value of digital assets lies in their consumption, providing access to AI models and algorithms via marketplaces can allow providers to market and monetize these APIs.
  • This ability to monetize AI by exposing it as an API is pushing editors to offer AI APIs as a product that can be consumed by developers everywhere to boost their digital services.