Produced by | Entrepreneurship Frontline

Art Editor | Xing Jing

Reviewed by | Song Wen

In the realm of enterprise digitalization, RPA (Robotic Process Automation) is synonymous with automated processes that take over repetitive and fixed workflows. Meanwhile, AI digital employees represent an intelligent operational upgrade from RPA, having once served as a pivotal tool for cost reduction and efficiency enhancement in finance, government, and corporate sectors.

However, with the advent of the AI boom, more sophisticated AI agents capable of autonomous decision-making have emerged, rapidly reshaping the competitive landscape in this field.

On August 21, 2026, Zhuhai Jinzhwei Artificial Intelligence Co., Ltd. (hereinafter referred to as “Jinzhwei”), a frontrunner in financial RPA, submitted its second application to list on the Hong Kong Stock Exchange.

This company, which rose to prominence on the wave of financial digitization, has deployed over 2 million AI digital employees over the past decade. It boasts the six major state-owned banks and 90% of securities firms as its core clientele, is supported by star investors such as Hillhouse and Qiming Venture Partners, and saw its valuation soar past 3 billion yuan in 2023.

Yet, beneath these impressive feats lies a stark reality: cumulative losses of 640 million yuan over three and a half years, persistent operating cash flow deficits, no new financing in three years, and a daunting 2.1 billion yuan in redeemable liabilities.

At this pivotal moment, as the RPA dividend wanes and AI agents emerge, can Jinzhwei, with its second listing bid, successfully navigate into the capital market?

1. A Banking Veteran Builds 2 Million AI “Workers,” Backed by Hillhouse

Jinzhwei’s founder, Liao Wanli, aged 54, graduated in computer science from Hangzhou Dianzi University (now Hangzhou Dianzi University of Technology) in 1993. He embarked on his career in technical development for financial trading systems at the Zhuhai Branch of the Agricultural Bank of China, where he spent eight years.

During his tenure, Liao observed the operational challenges securities firms faced before market opening: each morning, four staff members had to complete over 2,000 operational steps across seventy to eighty systems within a mere two hours. This arduous and repetitive manual process sparked his vision for process automation.

In 2005, Liao ventured into entrepreneurship in Guangdong, developing auditing software. In 2009, he returned to the financial industry, launching an automated market opening and closing system. In 2016, he founded Jinzhwei, targeting the financial industry’s RPA space and introducing the K-APA platform, which leverages AI digital employees to replace numerous repetitive and standardized back-office operations within banks and securities firms.

Over the ensuing decade, Jinzhwei capitalized on the financial industry’s digital transformation dividend.

As of June 30, 2026, Jinzhwei had deployed over 2 million AI digital employees, with its business spanning 270 banks, including the six major state-owned banks, serving over 130 securities firms (covering more than 90% of China’s securities firms), and more than 220 other major financial institutions, establishing itself as a prominent player in the financial RPA arena.

According to Frost & Sullivan, by revenue in 2025, Jinzhwei ranked first among Chinese AI digital employee solution providers, though its market share was a mere 3.3%, with the top five players collectively holding just over 10% of the market.

This indicates an extremely fragmented AI digital employee market, with no single company establishing a dominant monopoly. In a market with an annual size of just 7.81 billion yuan, a 3.3% share means Jinzhwei still has a long road ahead to gain true industry influence.

However, capital is always quick to sense opportunity. From 2020 to 2023, Jinzhwei successfully attracted investments from institutions including Hillhouse, Qiming Venture Partners, China Development Bank Manufacturing Transformation Fund, and Shunwei Capital. After its Series C funding in June 2023, the company’s post-money valuation reached 3.075 billion yuan.

(Image / Prospectus)

Prior to the IPO, Liao Wanli and his controlled entities collectively held 40.65% of Jinzhwei’s equity. Qiming Venture Partners held 12.26%, Kingdom Technology held 12.14%, Hillhouse held 7.34% through Shanghai Chenjun, China Development Bank Manufacturing Transformation Fund held 6.50%, and Shunwei Capital held 0.98%.

However, since 2023, Jinzhwei has not secured new investment for three consecutive years. With a “cold” primary market, can Jinzhwei successfully list on the capital market?

2. Cumulative Losses of 640 Million Yuan Over Three and a Half Years: Why the Persistent Cash Drain?

Financially, Jinzhwei’s operations present numerous contradictions.

It operates in an AI space growing at nearly 40% annually, yet its revenue growth rate is less than a quarter of the industry average; its gross margin consistently exceeds 50%, yet it incurs annual losses that widen each year.

From 2023 to 2025, Jinzhwei’s revenue was 217 million yuan, 243 million yuan, and 256 million yuan, respectively, with a three-year compound growth rate of 8.7%.

(Image / Prospectus)

During the same period, the Chinese enterprise-level AI solutions market grew from 35.9 billion yuan to 69.4 billion yuan, with a three-year compound growth rate of 39.04%.

In the first half of 2026, the company’s revenue grew by 41.66% year-on-year, showing a clear rebound, though still below the industry’s expected full-year growth rate of 46.97%.

(Image / Prospectus)

The slower revenue growth stems from the dual constraints of a project-based model and reliance on a single sector.

From 2023 to the first half of 2026 (hereinafter referred to as the “reporting period”), over 70% of Jinzhwei’s revenue came from project-based deliveries, with each revenue stream undergoing a lengthy process of “customer acquisition, bidding, development, delivery, and acceptance,” making revenue recognition highly dependent on customer acceptance timelines.

This model means growth is linear accumulation rather than exponential scaling, with a naturally low ceiling.

(Image / Prospectus)

More critically, after a decade of focus, the financial sector has reached its growth limits, with the company’s business covering the six major state-owned banks and over 90% of securities firms, leaving little room for incremental expansion in the existing market.

Although the company has been expanding into non-financial sectors such as government services, manufacturing, and energy, with non-financial revenue’s share rising from 14.2% in 2023 to 30.5% in 2025, overall revenue remains small and growth unstable, dropping to 23.2% in the first half of 2026, failing to effectively drive overall revenue acceleration.

To make matters worse, customer stickiness is declining in Jinzhwei’s core financial sector, with retention rates dropping from 74% in 2023 to 60% in the first half of 2026, and average deal sizes shrinking from 380,000 yuan to 150,000 yuan.

While the prospectus notes that projects in the first half are naturally smaller due to seasonality, the trend of declining average deal sizes persisted from 2023 to 2025, indicating challenges with retaining existing customers.

(Image / Prospectus)

Even more concerning to outsiders than slow revenue growth is the persistent profitability loss.

From 2023 to the first half of 2026, Jinzhwei’s net losses were 63 million yuan, 122 million yuan, 346 million yuan, and 109 million yuan, respectively, with losses widening each year and totaling 640 million yuan over three and a half years.

Notably, Jinzhwei’s gross margin has consistently remained above 50%. However, high margins coexist with persistent losses due to its cost structure.

From 2023 to 2025, the company’s annual sales expense ratio hovered around 29%, but in the first half of 2026, it surged to 92.1%, with R&D expense reaching 65.3%, combining to exceed 150% of revenue, creating short-term pressure where “more orders lead to higher short-term losses.”

The spike in expenses in the first half of 2026 resulted from three factors: first, share-based compensation expenses of 22.909 million yuan, accounting for nearly 40% of sales expenses, representing non-cash incentive costs; second, channel fees rising from 3.566 million yuan to 10.986 million yuan, a nearly threefold increase; third, project-based revenue being concentrated in the second half of the year, leaving a low revenue base in the first half, while fixed expenses like sales salaries occurred evenly throughout the year, passively amplifying the expense ratio.

(Image / Prospectus)

R&D is also under pressure. The simultaneous iteration of the traditional KAPA RPA product line and the new KiAgent intelligent agent product line, with dual product lines operating in parallel, has strained resource allocation and driven up overall R&D investment.

Under persistent losses, operating cash flow has deteriorated. The company’s collection period extended from 182 days in 2023 to 373 days in the first half of 2026, with accounts receivable recovery slowing significantly and operating cash flow remaining negative, intensifying liquidity pressure year by year.

(Image / Prospectus)

3. 2.1 Billion Yuan in Redeemable Liabilities Loom as the Agent Era Poses Transformation Challenges

Against a backdrop of persistent operating losses, a tightening primary market financing environment, and the fading effectiveness of old growth models, transitioning to AI agents represents a crucial path for Jinzhwei to sustain growth and break through industry bottlenecks.

However, transformation is never immediate; it is first and foremost a race against time.

While both RPA and intelligent agents aim to “let machines do the work,” their underlying logics belong to two different eras.

RPA “automates known processes,” requiring humans to define each step for machines to execute via scripts; intelligent agents “autonomously complete unknown goals,” needing only a task assignment for machines to independently plan paths, invoke tools, and make dynamic decisions.

With the full-scale emergence of AI agents in 2026, enterprise customers’ procurement logic is shifting from buying tools to buying AI capabilities.

Amid this sectoral evolution, Jinzhwei’s business structure weaknesses are gradually exposed.

For a long time, the company has relied heavily on the traditional RPA product, the K-APA solution, which accounted for over 93% of revenue from 2023 to 2025, serving as the absolute revenue mainstay.

Not until the first half of 2026, benefiting from the rollout of intelligent agent products, did Jinzhwei’s K-APA business share drop to 70.8%, with the Ki-Agent intelligent agent business growing rapidly, reaching 18.991 million yuan in revenue and accounting for 29.2%.

While the new business’s rapid growth offers fresh growth prospects, the intelligent agent business remains in its early deployment stage, with limited overall revenue volume and unable to replace traditional RPA as a second growth pillar in the short term.

(Image / Prospectus)

Moreover, the true intelligent agent era demands ongoing value delivery through “subscription, invocation, and revenue sharing,” rather than the one-time transaction model of “delivery, acceptance, and payment.” This requires breaking away from the traditional project-based model of “custom development for large clients, on-site implementation, and milestone-based acceptance.”

Whether Jinzhwei can navigate this fundamental shift from product form to business model will determine whether it can secure a “ticket” to the intelligent agent era.

And Jinzhwei has little time left for this transformation.

As of June 30, 2026, the company had just 71.891 million yuan in cash on hand, compared to redeemable liabilities of 2.126 billion yuan. If this Hong Kong IPO attempt is rejected, if it fails to pass a hearing within 18 months, or if it does not complete listing within six months after a hearing, relevant redemption clauses may be triggered.

For Jinzhwei, its second attempt to list on the Hong Kong Stock Exchange is more than just a repeat capital maneuver: successful listing would provide funds to buffer its AI agent business transformation; if the listing process stalls, the pressure to repay over 2.1 billion yuan in redeemable liabilities will become acute.

Looking back at Jinzhwei’s decade-long development, the deployment of 2 million AI digital employees and its comprehensive financial client base prove the company’s industry-leading scene delivery and implementation capabilities, which remain its core confidence in the sector.

However, in the new era of AI agents, the project experience and client resources accumulated during the RPA era are, after all, tickets from a bygone era. Whether they can adapt to the new journey of the AI agent era remains to be seen—and the funds and time needed for this validation are precisely what Jinzhwei lacks most right now.

*Note: The featured image in the article is from the Jiemian Gallery.