It is common for artificial intelligence (AI) experts from research and industry to talk about “agentic” AI as being at the heart of innovation. OpenAI’s Sam Altman said this almost a year ago during the company’s launch of an online store for custom GPTs and talked at length about how agents will take on much more complex tasks than co-pilots of today.
The next step beyond current agentic AI could be networks of agents that collaborate to complete more complex tasks, mostly without human supervision, according to Dharmesh Shah, co-founder and CTO of software maker Hubspot.
« I think over time we’ll see these agents collaborating with each other “, he declared to our colleagues at ZDNET.com.
« Agents are actually a step up from co-pilots. And I think both will have their place in the business landscape. What makes agents interesting is that they can set higher-order goals that typically involve multiple steps “, he continues.
Hubspot competes with companies like Salesforce to deploy various agents to help with customer relationship management (CRM) tasks, including sales, marketing, customer relationship management, and more.
To connect these agents, Hubspot promotes the idea of a network that acts as a marketplace for agents. At its annual user conference a week ago, in addition to a suite of AI offerings called Breeze, Hubspot unveiled a network for agents called agent.ai.
A LinkedIn for AI
The offering has more than 47,000 users and more than 1,700 builders have signed up to create their own agents, according to the company.
« Think of it as the number one professional network for agents, unlike LinkedIn, which is for humans “, he explains.
Thanks to the network, we can imagine teams of agents, made up of “mini-agents” and a supervising agent, explains Dharmesh Shah.
« Over time, as these agents develop, they will be able to, in some way, use each other. So an agent says, “I’m an agent who helps you research a company.”
Agents that we put together like Legos
This agent looks at public company transcripts, such as earnings conference calls. “ Then there is an agent who will view the website [de l’entreprise] and see how web traffic behaves and all these sorts of things, and it brings all this data together Dharmesh Shah developed : « So these kinds of mini global agents are used by some kind of higher order agent, so you kind of compose these agents like Legos and then you build higher order structures. »
The agents become “ digital teammates ”, according to him. The agents.ai network is a marketplace for finding which agents can perform which tasks, “ to see what their experience is, if they would be suitable or not » based on user feedback.
Past efforts to create collaborative “things,” such as the CORBA standard in the 1990s, were thwarted by interoperability, Dharmesh Shah recalls. This time around, the natural language capability of generative AI becomes the connective tissue for programming and assembling agents. “ We can interact with AI through natural language, right? Well, this also applies to agents », he adds.
« So to speak, an agent’s API is actually a natural language. You don’t need to learn this other language to be able to use these agents, whether you do it as a human or as another agent. This opens up a new level of interoperability that I think has always been difficult to achieve. »
A “paradigm shift” for CRM
For Dharmesh Shah, the increased importance of agentic AI and collaborative agent networks is a reinvention of CRM software.
“We saw a lot of innovation in the cloud era,” he explains, referring to the period 2006-2007 when cloud computing first emerged, “ which was the last major type of “transformation of all sectors”. Cloud CRMs have emerged and have now taken over almost every major type of CRM. »
The « paradigm shift » current in CRM will be led by AI. “ We are going to have a new type of paradigm now, which will be intelligent AI-based CRM »,
He believes the competitive battle in CRM between Hubspot, Salesforce and others will be about which “platforms” offer the best use of agents for users and developers.
« One of the reasons why I’m personally very excited about this sort of transition to AI is that we’re now going to see a whole new generation of developers who are going to be looking for a platform on which to develop their ideas. , saying, “Hey, I want to create something for marketing or sales ».
« Which CRM platform will they choose? I think we have an opportunity now to create this kind of new perception within the developer ecosystem to say, “Hey, now it’s not about web apps; agents are the new applications.” This is what we are going to create. So it’s exciting. »
