Build an AI for business
Bill Gates defines a platform as something where the economic value to everyone who uses it is greater than its value to the company that created it.
It is easy to see how Windows, through the various iterations of programming frameworks that have evolved over the years, is a platform on which commercial off-the-shelf software can be developed and hardware providers are able to sell standard PC equipment on which these applications are able to run. The Azure public cloud and Microsoft’s overall approach to software, is about trying to develop the next industry standard platform.
Microsoft’s next target is artificial intelligence. Referring to Github Copilot and Microsoft’s recent spate of Copilot announcements, the company’s CTO, Kevin Scott, used his keynote at Build 2023, to discuss why AI is changing application software.
Build is the company’s annual software developer’s conference. This year’s theme is focused on the idea of AI-infused applications. The idea of a conversational agent-like interface within software that helps end users do cognitively complex things can be applied across numerous IT systems. But building these things is hard, which is where Microsoft sees its role as a platform provider. “What makes platforms even greater than all of that value that they can potentially produce is it prevents folks from having to bear the burden of building very complicated things from the ground up,” Scott told the developer audience at Build 2023.
Building on what has come before it, Microsoft has adopted the OpenAI ChatGPT plug-in model to introduce the idea of a plug-in platform, where third-party developers can build their own data gathering plug-ins that other developers can incorporate into their applications.
Plugging AI into the enterprise
Among the third-party plug-ins announced include OpenTable and Wolfram Alpha. There are also plans to have plug-ins for Expedia, Instacart, Kayak, Klarna, Redfin and Zillow. But there is nothing to stop third-party developers creating their own copilot plug-ins such as in corporate IT, where an internal application needs to combine its data store with different streams of internal and external information sources. The plug-ins for the AI copilot potentially offer a way for a business to take advantage of large language models on the internet whilst keeping personally identifiable and commercially sensitive data within the corporate network.
Scott sees plug-ins as a way to augment AI foundational models. Referring to AI-infused applications that augment human users, he said: “You should have ways to build your application on top of this technology, even when the model itself isn’t complete or perfect.”
No doubt there will be rival AI plug-in platforms. But what we are seeing from Build 2023 is the start of the next iteration of AI, in the context of enterprise application augmentation.