Physical Address

304 North Cardinal St.
Dorchester Center, MA 02124

What’s next for agent AI? The founder of LangChain looks to the surrounding sponsors


Subscribe to our daily and weekly newsletters for the latest updates and content from the industry’s leading AI site. learn more


Agetic AI is the latest big trend in AI development, but what comes next?

Although artificial general intelligence (AGI) may still be some time in the future, there may be an intermediate step with a technique called ambient agents.

Results LangChainthe AI ​​pioneer of the agent, introduced the term “circular agents” on January 14. The technology that LangChain creates includes the eponymous open source LangChain framework that helps organizations to connect large language models (LLMs) together to achieve results. LangChain Inc. raised $24 million in cash in February 2024. The company also has a portfolio of commercial properties including LangSmith’s LLM Ops.

With traditional AI features, users often interact with LLM via voice commands to initiate action. Agentik AI generally refers to an LLM-powered machine that takes action on behalf of the user. The concept of peripheral agents allows the paradigm to move forward.

What are ambient agents?

Ambient agents are AI systems that run in the background, continuously monitoring streams of events and then triggering actions when appropriate, based on pre-set guidelines and user goals.

Although the term “surrounding agents” is new, the concept of wisdom aroundwhere the AI ​​is always listening, isn’t it. Amazon refers to its Alexa assistant technology as enabling ambient intelligence.

The goal of ancient agents is to replace repetitive tasks and improve user experience by having multiple agents running continuously, instead of the user having to call and interact with each one individually. This allows the user to focus on high-level tasks while the assistants work around the clock.

In order to help validate and improve the concept of circular assistants, LangChain has created a series of first user experiences, which manage emails, some social networks, to help users manage and respond when needed.

“I think the crowdfunding is really powerful and fun and cool,” Harrison Chase, cofounder and CEO of LangChain, told VentureBeat. “Environmental support is very powerful if there is a team of them doing things behind the scenes, you can develop yourself.”

The technology uses many open source solutions, and LangChain has not yet revealed how much it will charge for using new tools.

How virtual assistants work to advance the use of AI

Like many technical arts, the initial motivation for peripheral assistants was not to create new ideas, but to solve a real problem.

For Chase, the problem is one all too familiar to most of us: overflowing inboxes. Chase began his journey of creating a virtual assistant to solve the problems of email. Six months ago he started creating a plugin for his email.

Chase explained that the email service distributes its emails, and controls the test process. You don’t even have to edit it manually through its box, since the assistant takes care of it. Through his long-term use of the writing box, Chase was able to refine and improve his skills. He said that it started out imperfect, but with regular use and pain management, he was able to improve the agent’s performance.

To be clear, an email client is not some kind of easy way to manage emails. It’s a system that understands his email and helps him decide how to manage it.

A custom design with email support

The design of the Chase email client is very complex, including several components and language versions.

“It starts with a triage step that’s like an LLM and it’s very quick and a few samples are taken from the database,” Chase explained. “Then, if it’s confirmed that they’re trying to respond, it goes to the writing assistant.”

Chase also explained that the recorder has additional tools, including a calendar assistant:

“There’s an assistant I have specifically to coordinate with the calendar, because LLMs suck at dates,” Chase said. “So I had to have a dedicated assistant just to keep up with the calendar.”

After those answers are created, Chase said there is an additional LLM phone that rewrites the answer to ensure proper tone and formatting.

“I found that having an LLM trying to dial up all these devices and create an email and type in the correct wording was difficult, so I have a step up in sound,” Chase said.

Inbox as a way to control and manage agents

The most important part of the environment is having control and transparency over what the agent is doing.

Chase noted that in the initial setup, they only had support messages via Slack, but that became a bit confusing.

Instead, LangChain created a new user interface, the supporter login box, specifically to communicate with surrounding supporters.

Screenshot of the LangChain support box. Credit: VentureBeat

The system shows all the communication channels between users and agents and makes it easy to follow the action.

How to create a circular space

LangChain is first a tool for developers and will be a tool for building and deploying virtual assistants now.

Any developer can use LangChain’s open source technology to create a blockchain, although additional tools can make the process easier. Chase explained that the help box he created is visible at the top of the LangGraph platform. LangGraph is an open source framework for architecture assistants that provides a framework for working with primitives.

On top of that, LangChain is using its trading platform LangSmith, which provides features and analysis for sponsors. This allows developers to equip design assistants with the necessary monitoring and evaluation tools to ensure they are performing as expected.

Ambient agents: A path to universal intelligence

Chase is optimistic that the concept of peripheral assistants will work with manufacturers in the coming months and years.

Surrounding agents bring the prospect of autonomy to AI, enabling it to monitor events and act intelligently. Chase still expects that there will be a need to keep people on track as part of the surrounding events. But people only need to confirm and confirm actions, instead of knowing what to do.

“I think it’s part of helping to use more common sense,” Chase said.

Chase felt that it was possible that true AGI would result from a change in thinking. That said, good design work of examples is where the concept of circular assistants will bring value.

“There is a lot of work to be done using the models, even if they are very smart,” Chase said. “I think that the way to communicate with them will be the best way to use this kind of intelligence.”

The latest version of the email client is available. LangChain is releasing a new social media platform today, and will make its supporter inbox available on Thursday, January 16.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *