Aiming to become a digital labor production factory, "Future Intelligence" has completed a Pre-A round of financing.

Text by | Wang Xinyi

Edited by | Deng Yongyi

36Kr learned that AutoAgents.ai recently completed a Pre-A round of financing. The new investors include Fanchuang Capital, Zhongguancun Capital, and Tanyuan Capital. Existing shareholders Orient Securities Innovation and Lingge Venture Capital also participated in the follow-on investment. The funds from this round of financing will be mainly used for computing power investment, team expansion, and the ecological construction and operation of new products.

AutoAgents.ai was founded in June 2023 and focuses on empowering knowledge workers with agent technology. The core team members come from Alibaba DAMO Academy, Tencent, ByteDance, and Google. Yang Jinsong, the founder and CEO, once served as the product/commercialization director at Alibaba DAMO Academy. He was also the person in charge of ByteDance's Feishu AI and the head of Amazon AWS's aPaaS platform. He led the launch of Alibaba Lingjie and Tongyi - Alicemind and has many years of experience in AI product and commercialization.

As one of the core products, the enterprise-level agent building platform "Lingda" aims to address the core requirements of enterprises when deploying and applying large models: data security and privacy, hierarchical permission management, complex system integration, and the delivery stability of agents in real business scenarios.

Since its inception, "Lingda" has focused on enterprise customers, especially those in industries such as power, finance, and manufacturing, which have extremely high requirements for stability and compliance.

Back in 2023, when the "hundred-model battle" was in full swing and most entrepreneurs and investors flocked to the large model track, Yang Jinsong made a different decision - not to develop large models but to focus on agents.

His experience in personally training large models at Alibaba DAMO Academy made him realize that large models will become infrastructure but not the final consumption form. Enterprises need a system that can deliver results, which is exactly what agents can solve.

Lingda is positioned as a low-code AI agent development platform for business personnel. Different from canvas-style products like Coze and Dify, Lingda pays more attention to the needs of non-technical users and aims to lower the threshold for business personnel to build and use agents.

In terms of product capabilities, Lingda provides more than 20 standard module nodes, covering user inquiries, AI conversations, information classification and extraction, knowledge base search, document review, image recognition, database query, etc. It also supports Text2Agent natural language generation workflows and the Skills engine, transferring the cumbersome work that originally required an IT team or an external implementation team to people closer to the business site.

At the architectural level, Lingda adopts a "cloud + edge" design: the cloud side accumulates and precipitates the core knowledge assets within the organization, including agents, exclusive skills, and work contexts; the edge side allows employees to directly and securely access the enterprise's full knowledge base through local clients and precipitate new skills in daily high-frequency business interactions, which are finally fed back to the cloud knowledge base.

In the B2B market, AutoAgents.ai has achieved large-scale implementation of benchmark scenarios in multiple vertical industries. Starting from the contract review project for East China Grid, a seed customer, Lingda has covered more than 20 grid customers with a 100% renewal rate, and its market share ranks among the top in the agent product field.

In 2024, AutoAgents.ai achieved revenues of several million yuan, and in 2025, the revenue increased by four times. The overall revenue comes from multiple industries such as power, finance, and manufacturing. The goal for this year (2026) is to reach 100 million yuan.

After accumulating rich industry experience and typical templates with large B2B customers, AutoAgents.ai recently officially launched another heavyweight strategic new product - the AI Digital Expert Marketplace "Daidai".

If "Lingda" is a production factory for AI digital labor, providing enterprises with an underlying platform for building and managing agents; then "Daidai" is an agent employment platform that digitizes the in-depth knowledge of human experts and packages them into digital employees that can directly deliver results. Users can directly "hire a digital expert" on the platform and pay based on the results.

Since this year, Agent tools such as OpenClaw have been popular for a while and then quickly cooled down. Yang Jinsong believes that in the future, most users should "use" agents rather than "maintain" them, and Daidai is the product of this concept.

The Daidai platform currently offers two modes: the application mode is used for efficient tasks in specific scenarios such as customs declaration and tax filing; the expert mode deeply undertakes complex job functions such as AI video production, e-commerce marketing material generation, self-media operation, and investment due diligence.

It is reported that the Daidai team has currently reached cooperation with nearly a hundred human experts to help them digitize their capabilities for knowledge monetization. With the verification of early seed customers, Daidai has achieved an ARR potential of over 10 million yuan.

The parallel development of "Lingda" and "Daidai" constitutes the unique "Harness Engineering" collaborative flywheel of AutoAgents.ai. The connotation of this concept is to use the massive task data generated from real scenarios to refine every decision of agents in reverse.

The two product lines will form a collaborative closed-loop - the user data and task trajectories generated by Daidai are fed back in real-time to iterate the underlying model and agent capabilities of Lingda; the stronger agent components evolved on Lingda are then re-listed on Daidai for users to use. Driven by this dual-wheel mechanism, the task success rate of the digital employees produced by the platform has increased significantly from the initial 72% to 91%.

Looking at the customer profiles, the customers of "Lingda" are mainly concentrated in industries such as energy, finance, and manufacturing, such as State Grid, large banks, and securities firms. "Daidai" initially targets small B2B and C2C customers. This year, Lingda hopes to promote the product to the stage of large-scale replication in the industry and quickly spread the existing methodologies and tools to institutions such as securities firms and banks.

Yang Jinsong said that the essence of the agent business is "exchanging computing power for human resources." Taking an enterprise customer in East China as an example, after using AutoAgents' solution, the labor cost of a certain business has been reduced to one-tenth of the original.

"The current basic model capabilities are far enough to meet customer needs. The problem lies in whether it can be deeply integrated into the scenario and stably delivered," Yang Jinsong told "Intelligent Emergence". As the Token cost decreases, the ROI of agents will show an extremely high and irreversible trend.

As AI takes on more and more basic work, the talent profile of organizations is also changing. AutoAgents.ai will achieve the optimal delivery level in specific vertical tasks based on the long-term accumulated scenarios and data.

Yang Jinsong predicts that as the agent capabilities continue to improve, primary tasks within the organization may be completed by AI, and enterprises will value employees' high-level abilities more - in-depth understanding of the business, commanding AI, checking AI results, aesthetic judgment, and how to use agents to optimize existing business processes.