Microsoft CEO Satya Nadella: The era of AI agents will disrupt SaaS applications.
As artificial intelligence rapidly evolves, Microsoft CEO Satya Nadella believes that SaaS (Software as a Service) applications will be disrupted in the era of AI agents. In an interview, Nadella stated, “SaaS applications may be disrupted in the era of AI agents.” He explained, “They are essentially just CRUD (Create, Read, Update, Delete) databases with some business logic added.”
Nadella further elaborated that this business logic will all be transferred to AI agents. These agents will be multi-repository CRUDs and won’t distinguish what’s in the backend. They will update multiple databases, and all the logic will be concentrated in the AI layer. Once the AI layer becomes the place where all logic is executed, people will start replacing the back end. This means that the logic layer of business programs can be orchestrated by AI and AI agents.
Although there are many large and valuable SaaS applications in existence today, essentially, what they do is very simple. Most SaaS applications have a database to store data and then perform CRUD operations. For example, a SaaS application like Notion is essentially a database where users can store data, update it, and display it in a convenient format. Nadella seems to be suggesting that the intelligent part of database interaction can soon be handled by AI agents, which may make SaaS applications redundant. Moreover, these AI agents can interact with multiple databases simultaneously. For instance, an AI agent can read a memo containing an email address from the Notion database and then send an email through Mailgun. Since AI agents will control the interaction with the database, the value of SaaS applications like Notion and Mailgun will be reduced to simply maintaining a database, while the value will be provided by the AI layer above.
However, this is not the only threat facing SaaS applications. Thanks to advances in AI, it is now much easier to write simple applications than before. It is even possible for organizations or individuals to rebuild large SaaS applications on their own. In fact, Chamath Palihapitiya has founded a company called 80:20 that uses AI to achieve 80% of the functionality of SaaS applications at 20% of the cost. It seems that AI will not only reduce the price of SaaS applications but may also change the way they are used through the AI layer running on top.
Nadella said that Microsoft has achieved quite good results in dynamics, backends, and agent usage. They will actively promote and try to integrate all of this together, whether in the field of customer service or other fields. By the way, another growing trend is that not only CRM but also finance and operations are increasingly shifting to AI-native applications.
This means that the logic layer of business programs can be orchestrated by AI and AI agents. In other words, the transition from Copilot to agents and then to business programs should be very seamless. Now, one might even ask, why do we still need Excel? Interestingly, one of the things that excites Nadella the most is that the combination of Excel and Python is like the combination of GitHub and Copilot.
Essentially, when you have a tool like Excel, you should open Excel, start Copilot, and start playing around because it’s no longer just simple number crunching. It’s like a data analyst. It’s no longer just about helping you understand existing numbers. It will make plans for you, just like GitHub Copilot creates and executes plans.
This is like a data analyst using Excel as a row-column visualization tool for analysis. Copilot will use Excel as a tool and take advantage of all its operating space because it can generate code and has a built-in Python interpreter. In fact, this is a great way to redefine Excel.
To some extent, it can be said that an entire Excel can be generated because after all, it has a code interpreter. So anything can be generated. Therefore, Nadella believes there will be a disruptive change. But currently, when dealing with Microsoft’s M365 products, the approach is to first establish a Copilot as the interface layer of organizational AI, obtain all agents, including Microsoft’s agents.
It can be said that Excel is an agent for Copilot, and Word is also an agent. They are specialized canvases. When working on legal documents, you can bring it to Pages, then to Word, and finally to Excel. Copilot assists throughout the entire process. This is a whole new way of working.