Sequoia Capital's latest insights: 2025 will be the year of the Agent! AI entities enter the era of "collective collaboration."

Sequoia partner Konstantine Buhler recently gave an exclusive interview to Bloomberg, where he emphasized that the current hot topic in AI is "agents," which are AI systems capable of completing specific tasks.

He predicts that from 2025 onwards, AI agents will evolve from single agents to a "collaborative network" model, where multiple agents work together and even compete to accomplish more complex tasks.

Additionally, he pointed out that key sectors such as healthcare and education will become the next important battleground for AI technology, with the potential to reduce costs and improve efficiency, thereby driving social progress.

Here are the main points from the interview:

Host

Are we still excited, Konstantine?

Konstantine Buhler

We are indeed in a cycle of AI hype, but our focus is not just on the hype of 2024 or 2025, but rather on long-term development.

AI is a very important and large-scale macro trend that can be compared to the Industrial Revolution. This is also a field we have been investing in for decades and will continue to invest in for decades to come, regardless of how the hype changes.

Host

You once wrote an article mentioning how organizations like Gartner view the hype cycle, from the initial excitement to the later disappointment phase—the so-called "trough of disillusionment."

What stage are we currently in regarding consumer adoption rates and the actual progress of converting AI investments into returns?

Konstantine Buhler

You are absolutely correct. Let me give you a few examples. AI evolves over time. When I was an AI researcher, which was a long time ago, we had just begun to explore natural language processing and understanding, as well as the initial explorations of computer vision.

By 2022 and 2023, the focus shifted to large language models for answering questions and the birth of modern AI.

2024 will be the year of "agents." An agent is an AI that can complete tasks. When you ask it to do something, it not only tells you how to do it but also does it for you directly.

For example, one of the companies in our portfolio, Expo, is an agent company. The founder of Expo is the creator of GitHub Co-Pilot, and they focus on developing agents for the cybersecurity field, which is very challenging and critical.

In just a few months, they developed an AI that can actually surpass some of the top penetration testers in the world.

This AI enters and attacks a website with the permission of the website provider, finds vulnerabilities, and reports the results to the user, indicating that the entire task has been completed. This is the focus of AI in 2024—agents As for 2025 and beyond, it is likely to shift towards the collective collaboration of agents, that is, a network of multiple agents that cooperate and even compete with each other to achieve certain goals.

Host

The collective collaboration of agents sounds like a group of companies discussing agents. Salesforce's latest financial report also focuses on "Agent Force."

At the same time, one of your portfolio companies, Rocks, is also exploring agents. Will we see how these agents influence each other, cooperate, or experience some successes and failures?

Konstantine Buhler

Absolutely. The reason people are so excited about agents is that the potential for AI to complete tasks is enormous—not all tasks, but many tasks can be achieved.

This is highly impactful for us. You just mentioned Rocks, which is a great example. The agents developed by this company not only complete tasks but also integrate humans into the process.

Rocks does not aim to replace people but rather says that you are an excellent salesperson with a good relationship with buyers.

We are not looking to sever that connection but to enhance your capabilities through agents. This agent will handle all the work and research for you, helping you prepare for the next meeting.

A practical example is that they initially developed a technology for automatically writing emails, even capable of sending them automatically. However, they later removed the automatic sending feature and found that performance improved by 333 times when human salespeople were included in the process.

Therefore, some companies will develop technologies that are entirely agent-driven, such as Expo in the cybersecurity field; while others will use agents to assist humans, like in the case of Rocks.

Host

You are a visionary who always releases trend forecasts. Predict for us, how many types of agent products will we need? How many public and private companies will be involved? How many opportunities are there for companies like Rocks?

Konstantine Buhler

The era of AI has just begun. The key limitation lies in our creativity. The scale of opportunities for AI is enormous. I will give two examples that are crucial for all of us: healthcare and education.

The costs in these two fields have been rising for decades, primarily due to high service costs. Now, AI can actually complete some tasks.

I believe that in the coming decades—not just next year—AI can become a hero in these industries, reducing costs and helping us achieve a more educated and healthier society. Therefore, the potential of AI, agents, and AI networks is almost limitless.

Source: Youxin Newin, original title: "In-depth | Sequoia Capital's Latest Insights: 2025 Will Be the Year of Agents! AI Agents Enter the Era of 'Collective Collaboration'"

Risk Warning and Disclaimer The market has risks, and investment should be cautious. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this is at one's own risk