At Google’s cloud event, they announced a new protocol for AI agents to speak to other AI agents. Agent2Agent protocol becomes Change how the web works. This is not a dark possibility that can happen in the future. It’s here now!
In a recent interview, Deepmind -CEO Demis Hassabis, whose company is a leading research laboratory in Google, said that he is of the opinion that we will end with an internet where Agents speak to other agentsNegotiate things between yourself and then give us back the results. This means that the structure of the web and the way we use it change.
Google says, “Every company will soon rely on multi-agent systems.“These agents are intelligent systems that act in our name, planning and memory in our name. We will soon have personnel agents who are looking for helpful agents on the Internet and interact with them.
The SEO will probably shift from the optimization for search engines to optimize agents to other agents.
In this article I will discuss some of the things that are announced in Google Cloud – Agentspace, Customer Engagement Suite, Agent Development Kit and the most interesting announcement, Agent2Agent Protocol.
Agent space
Agentspace is a way for your company to access your documents and data with AI and to do things. Currently, Access to Agentspace Is only invited. I was told that you have to pay 100 seats to have access, even though I expect this will probably change.
Here is a clip from the next event of the cloud that shows how Agentspace works. Each one in the company has their own collection of agents who do things for them. These agents communicate with company -wide agents This can synthesize company data, create reports, draw from searching and much more.
They gave an example of a company that had an onboarding agent to welcome new employees. This onboarding agent can communicate with an HR agent or possibly with a salary accounting agent.
Customer loyalty suite
This clip of Google’s customer loyalty suite from Google Is a must. In the demo, the man interacts with a chat bot from a garden business. It looks like a typical chat bot except that it communicates with voice. The voice sounds very real! He asks what type of floor he should buy. The agent asks to see his plants and sends a video request. “She” can then see that he plants petunia. She says that the soil he had added to his car was the wrong way and offers to put the right way in his carts.
They also discuss that a landscape designer plants for him. He asks about a 50%discount. The agent says she first has to inquire with her supervisor. She then sends a message to a human call center manager who authorizes a 20% discount, which she then offers to the customer. You will then plan an appointment.
The Customer Engagement Suite from Google changes the way we interact with companies. I assume that this suite will not only interact with customers, but will probably also give us AI -controlled insights into the interaction of customers with our business.
Agent development kit
Google Agent development kit is an open source framework that makes it very easy to build agents. You can do this in less than 100 lines of Python. Even if you don’t know how to code, I would recommend spending time to understand what this code is doing. You can insert it in Chatgpt, Gemini or Claude and ask for help with the implementation. If you are brand new in the coding, try such input requests,
“I have never coded before, but I would like to learn how to create an agent by following this documentation (inserting in the ADK documentation). Give me through everything I have to do. Take it a step in a row because I am a beginner. ”
If every company will have several agents, then There will soon be an incredible demand for people who understand how to build agents And what we talk about next How agents communicate with each other.
One of the most interesting things I found in the documentation was the list of example remedies in the Agent Garden. This includes a data science agent, a customer service employee and the most interesting “one”Brand search optimization”Agent. This agent will determine important keywords for a brand that analyze search results, spy on competitive strategies, compare differences in titles day, descriptions and more and propose implementable recommendations for improvement.
I would encourage you to put time to try to build this type of agent.
It will be important because The web will move from users who use search engines to find information on agents that find other agents with whom they can communicate.
Agent2Agent protocol
Google terminated a new protocol called the an Agent2Agent (A2A) protocol This enables agents to communicate with each other, regardless of which LLM they were used to create them. The “agent card”, A is central to this protocol Json-formatted advertising Where agents explain their skills and enable other agents to discover them.
Agents will certainly communicate with each other and the output of their task, referred to as “artefact”.
How could A2A change the way the website works?
I believe that in the development of this protocol, a lot of what is currently happening when people enter keywords and search for websites Agents who are looking for other agents This is probably helpful.
For example, say I want to buy a new DSLR camera. Today I could search for “Best DSLR camera”, search the manufacturer’s websites and read some reviews. In a world powered by agents, I would ask my LLM assistant (perhaps Gemini or Chatgpt or even a private one that I create) to carry out this research for me. Instead of the agent, who is on the web, he is looking for agent cards that are exchanged for the information to increase the purchase of cameras. My agent can receive specifications from the agents of Canon, Nikon and others. It can find a helpful evaluation agent created by a professional photographer who checked cameras. It will know my preferences and preferences and ask specific questions from these agents so that it can give me the most helpful answer.
Now there is an obvious question here and that is What is the motivation for the camera blogger to create content If my agent does not show any ads or click on affiliate links from his website. Google announced a AI agent market. I could see a scenario in which people who create really useful and helpful content have agents for whom it is worth paying.
For example, for me there is what incentive I can write about AI News In a world in which your agent can simply curate a list of interesting messages for you from the last search results? If my AI messages also contain helpful requests, thoughts of community members and tips and strategy from me (it does!), This is an agent for which people would likely pay for.
I think there will be even more incentive to create content in an agent powered ecosystem. Only those who can create content that is really worth paying will probably be successful. Perhaps agents begin to learn which agents find others helpful. I can see a world in which we work to optimize agents and agent cards to look most attractive for other agents.
Podcast
I have a lot of thoughts about this new protocol and how agents will change our lives! Here is mine Podcast This fits this article, in which I talk more about how certain types of companies are probably influenced and what opportunities are ahead of us.
https://www.youtube.com/watch?v=BHRSFH3E3QI
YouTube | Apple podcasts | Spotify | Google podcasts | Soundcloud
And if you like that, you will love my loved ones Newsletter – It’s free! I work very hard every week to curate the most interesting and helpful news about AI because it changes the online world.
Marie
Further reading:
The Vertex AI offers new opportunities to build and manage multi-agent systems
Customer loyalty suite with Google Ai
Agent Development Kit: Just make it possible to create multi-agent applications
Google’s Agent Development Kit Quickstart Guide – These are really simple instructions for building an agent in Python. You can copy and insert this documentation into your preferred LLM and should be able to create an employment agent.