Emergence AI’s new system automatically creates AI agents rapidly in realtime based on the work at hand

Another day, another announcement about AI agents.

Known as a market Research Report With the big technology trends in 2025 (especially in enterprises), we can’t seem to be able to go beyond 12 hours or so without other ways to make, orchestrate (link together) or otherwise optimize specially built AI tools and workflows.

However AI appearsa startup founded by a former IBM research veteran It debuted in the second half of last year, cross-platform AI agent orchestration frameworkaccompanied by novelty in everything else: an AI agent creation platform that lets human users specify the work they are trying to accomplish with text prompts and then hand it over to the AI ​​model to create the agents they think are necessary to do the above.

This new system is actually a codeless, natural language, AI-driven multi-proxy builder and works in real time. Advent AI describes it as a milestone in recursive intelligence, aiming to simplify and accelerate complex data workflows for enterprise users.

“Recursive intelligence paves the way for agents to create agents,” said Satya Nitta, co-founder and CEO of AI emergence. “Our system allows creativity and intelligence to expand smoothly without human bottlenecks, but always within the boundaries defined by humans.”

The platform is designed to evaluate incoming tasks, check their existing proxy registry, and generate new proxy autonomously if necessary to meet specific enterprise needs. It can also proactively create proxy variants to predict related tasks and expand its problem-solving capabilities over time.

According to Nitta, the band’s architecture enables a completely new level of autonomy in enterprise automation. “Our orchestrator stitches multiple agents together to create uncoded multi-agent systems. If it doesn’t have a proxy for tasks, it will automate a single and even self-play by creating a new proxy itself, and can even self-play to learn related tasks.”

A brief demonstration presented in a video call last week seemed impressive, with NITTA showing how simple text instructions for AI classification triggering a wave of new agents, showing a wave of new agents on the visual timeline, showing the colored dots each agent represents in each column to indicate the category of work it is designed to help it start.

Nitta also said that users can stop and interfere with the process at any time and communicate other text instructions at any time.

Bring agent coding to enterprise workflow

Technologies that emerged focus on automating data-centric enterprise workflows such as ETL pipeline creation, data migration, transformation and analysis. The platform’s agents are equipped with proxy rings through planning, verification and self-play, long-term memory and self-improvement capabilities. This allows the system to not only complete a single task, but also understand and browse the surrounding task space of nearby use cases.

“We’re in a weird time in the development of technology and society. We have AI joining the conference now,” Nitta said. “But beyond that, one of the most exciting things that have happened in AI in the past two years is that large language models are making code. They’re getting better, but they’re systems of probability. That code may not always be perfect, and they don’t execute, validate or correct.”

Ai ai’s platform attempts to fill this gap by integrating the code generation capabilities of large language models with autonomous proxy technology. “We combine the code generation capabilities of LLMS with autonomous proxy technology,” Nitta added. “Proxy coding has a huge impact and will be the story of next year and in the coming years. The disruption is far-reaching.”

The emergence of AI emphasizes the ability of the platform to integrate with leading AI models, e.g. Openai’s GPT-4O and GPT-4.5,,,,, Claude of Humanity 3.7 Sonnetsand Yuan Camel 3.3and frameworks such as Langchain, Crew AI and Microsoft Autogen.

The focus is interoperability – allowing businesses to bring their own models and third-party agents to the platform.

Extended multi-proxy functionality

With the current version, the platform extends to include connector proxy and data and text smart proxy, allowing businesses to build more complex systems without writing manual code.

The ability of a performer to assess his own limitations and act is crucial to the approach that emerges.

“A very extraordinary thing that is happening is that when a new task appears, the orchestrator can solve the task by checking the registry of the existing agent,” Nitta said. “If it can’t, it creates a new agent and registers.”

He added that the process is not only reactive, but also generated. “The orchestrator is not just creating an agent; it is creating a goal for itself. It says, ‘I can’t solve this task, so I’m going to create a goal for making a new agent.” That’s what’s really exciting. ”

BET, so you don’t worry about the orchestrator getting out of control and creating also Research on many unnecessary custom agents on their platform shows that it has been designed and successfully performed the additional requirement to win the number of agents as it gets closer to completing the task, thus adding agents, making the agent more suitable for their internal registry to implement their internal registry. your Business, then check it out before creating any new business.

Prioritize security, verification and human surveillance

To maintain supervision and ensure responsible use, AI incorporates several security and compliance features. These include guardrails and access controls, verification titles that evaluate agent performance, and human supervision that validates critical decisions.

Nitta stressed that human supervision remains a key component of the platform. “The people in the loop are still important,” he said. “You need to verify that a multi-agent system or a new agent is performing the tasks you want and moving in the right direction.” The company has built a platform with clear checkpoints and verification layers to ensure that businesses retain control and visibility during automation.

Although pricing information has not been disclosed, AI invites businesses to contact them directly for access and pricing details. Additionally, the company plans to make further updates in May 2025 that will expand the platform’s ability to support container deployments in any cloud environment and create extended proxy through self-play.

Outlook: Expanding Enterprise Automation

The advent of AI is headquartered in New York with offices in California, Spain and India. The company’s leadership and engineering teams include alumni from the AI ​​research labs and technology teams at IBM Research, Google Brain, Allen AI, Amazon and Meta.

AIS AI describes its work as still in its early stages, but believes its recursive intelligent approach can disband new possibilities for enterprise automation and ultimately make AI-powered systems more extensive.

“We think the proxy layer will always be necessary,” Nitta said. “Even if the model becomes stronger, it’s very difficult to generalize in action space. People like us have enough room to advance this over the next decade.”