Emergence thinks it can crack the AI ​​agent code

Another generative AI venture has raised a massive amount of funding. And, like other ventures before it, this one is hoping for the moon.

Emergence, whose co-founders include Satya Nitta, the former head of global AI solutions at IBM’s research division, emerged from stealth on Monday with $97.2 million in funding from Learn Capital and credit lines totaling more than $100 million. Emergence claims to build an “agent-based” system that can perform many of the tasks typically handled by knowledge workers, in part by routing these tasks to first- and third-party generative AI models such as OpenAI’s GPT-4o.

“At Emergence we are working on many aspects of the growing field of generative AI agents,” Emergence CEO Nitta told TechCrunch. “In our R&D labs we are advancing the science of agentic systems and looking at it from a ‘first principles’ perspective. This includes critical AI functions like planning and reasoning, as well as self-improvement in agents.”

Nitta says the idea for Emergence came shortly after he co-founded Merlin Mind, which builds education-oriented virtual assistants. He realized that some of the techniques he developed at Merlin could be applied to automating workstation software and Web apps.

So Nitta recruited his former IBM colleagues Ravi Kocku and Sharad Sundararajan to launch Emergence, with the goal of “advancing the science and development of AI agents,” in Nitta’s words.

“Current generative AI models, while powerful at understanding language, lag behind in the advanced planning and reasoning capabilities needed for more complex automation tasks, which is what agents are all about,” Nitta said. “This is what Emergence specializes in.”

Emergence has a very ambitious roadmap that includes a project called Agent E, which seeks to automate tasks like filling out forms, searching for products on online marketplaces, and navigating streaming services like Netflix. An early form of Agent E is already available, trained on a mix of synthetic and human-annotated data. But Emergence’s first finished product is what Nitta describes as an “orchestrator” agent.

This orchestrator, the open source Monday, doesn’t perform any tasks itself. Rather, it acts as a kind of automatic model switcher for workflow automation. Taking into account things like the model’s capabilities and cost of use (if it’s third-party), the orchestrator considers the task to be performed – for example, writing an email – then chooses a model from a list curated by the developer to complete that task.

“Developers can add the proper guardrails, use multiple models for their workflows and applications, and switch to the latest open source or common models on demand, without worrying about issues like cost, quick migration, or availability,” Nitta said.

Emergence’s Orchestrator sounds similar in concept to AI startup Martian’s Model Router, which takes a prompt intended for an AI model and automatically routes it to different models based on things like uptime and features. Another startup, Cradle, offers a more basic model-routing solution driven by hard-coded rules.

Nitta doesn’t deny the similarities. But he very subtly suggests that Emergence’s model-routing technology is more reliable than others; he also says it offers additional configuration features like a manual model selector, API management, and a cost overview dashboard.

“Our Orchestrator Agent is built with a deep understanding of the scalability, robustness, and availability that enterprise systems require, and is based on the decades of experience our team has in building some of the most highly scaled AI deployments in the world,” he added.

Emergence intends to monetize Orchestrator in the coming weeks with a hosted, premium version available via API. But that’s just one part of the company’s grand plan to build a platform that can, among other things, process claims and documents, manage IT systems and integrate with customer relationship management systems like Salesforce and Zendesk to prioritize customer inquiries.

To that end, Emergence said it has formed strategic partnerships with Samsung and touch display company Newline Interactive — both of which are existing Merlin Mind customers, which doesn’t seem to be a coincidence — to integrate Emergence’s technology into future products.

What specific products and when can we expect to see them? Samsung’s WAD interactive displays and Newline’s Q and Q Pro series displays, Nitta said, but he didn’t have an answer to the second question, meaning it’s very early days.

There’s no denying that AI agents are all the rage right now. Generative AI powerhouses OpenAI and Anthropic are developing task-performing agentic products, as are big tech companies like Google and Amazon.

But it’s not clear what makes Emergence unique, other than the large amount of cash it receives in the early stages.

TechCrunch recently covered another AI agent startup, Orbi, which had a similar sales pitch: AI agents trained to work in a range of desktop software. Adept was also developing similar technology, but despite raising over $415 million now reportedly finds itself on the verge of a bailout from either Microsoft or Meta.

Emergence is positioning itself as the most R&D-heavy: the “OpenAI of agents,” if you will, with a research lab dedicated to investigating how agents can plan, reason, and improve themselves. And it’s drawing from an impressive talent pool; many of its researchers and software engineers are from Google, Meta, Microsoft, Amazon, and the Allen Institute for AI.

Nitta says Emergence’s guiding light will be to prioritize openly available work while building paid services on top of its research, a playbook borrowed from the software-as-a-service sector. He claims that early versions of Emergence’s services are already being used by thousands of people.

“We believe our work will become the foundation for automating a variety of enterprise workflows in the future,” Nitta said.

I’m skeptical, but I’m not sure Emergence’s 50-person team can outcompete the rest of the players in the generative AI space – nor will it be able to solve the broader technical challenges that plague generative AI, such as hallucinations and the huge cost of developing models. Cognition Labs’ Devin, one of the best-performing agents for building and deploying software, managed only a 14% success rate on a benchmark test measuring its ability to resolve issues on GitHub. Clearly there’s a lot of work to be done to get to the point where agents can juggle complex processes without any oversight.

Emergence has the capital to try now. But that may not be the case in the future as VCs and businesses are expressing doubts about the ROI path of generative AI technology.

Nitta, displaying the confidence of a man whose startup just raised $100 million, insisted that Emergence was well positioned for success.

“Emergence is resilient due to its focus on solving fundamental AI infrastructure problems, delivering clear and immediate ROI for enterprises,” he added. “Our open-core business model, combined with premium services, ensures a stable revenue stream while fostering a growing community of developers and early adopters.”

we will see soon.