Result as a Service: How to Build the RAAS Moat for Industrial Internet?

[Ebrun Original] Recently, Sequoia Capital made a prediction that 'Result as a Service (RAAS) will be the trend for the next decade.' Amidst the commentary on this capital narrative and industrial reality, a more fundamental question has surfaced: What is the true core competency of RAAS? The answer may not lie in the AI technology itself, but in an enterprise's ability to harness data as a factor of production. This includes both the application capability to deeply integrate AI with data resources and the governance capability to continuously manage these resources and build unique barriers.

01 Which is More Important? The Perspectives of Capital Narrative and Industrial Reality

In its report titled 'Services: The New Software,' Sequoia Capital made a somewhat disruptive assertion: The era of simply creating AI tools is over. The world's next trillion-dollar tech giant will emerge from the AI services sector, a service-oriented enterprise disguised as software that directly delivers results.

The report's core argument hinges on a comparative data point: for every dollar enterprises spend on software tools, they spend six dollars on manpower and business services. If current AI technology has reached a stage of market maturity, then the shift from 'selling tools' to 'selling results' is inevitable. This also means AI companies will move from the relatively narrow market for technical tools into a trillion-dollar service market with double the scale.

This argument, due to its logical consistency, quickly ignited discussions in the venture capital and startup circles, with many considering it the 'standard answer' for the future. However, controversy followed. Some industry observers astutely pointed out a fundamental flaw in Sequoia's vision of service 'Autopilot full-automatic result delivery' – real-world service scenarios do not involve purely rule-based, zero-variability processes. The basic nature of 'service' differs from that of a product. AI will inevitably make mistakes when handling non-standardized tasks, requiring human backup. When AI companies transition from technology to services, they will inevitably see asset-light businesses downgrade into service models with heavy human delivery, lacking the genes for scalability.

Currently, most AI projects in the market claiming 'result delivery' can be categorized into three types: RPA rebranding, lightweight AI-assisted tools, or digitized upgrades of traditional outsourcing, none of which achieve genuine full-automatic delivery.

So, is RAAS a disruptive business model innovation or an unrealizable capital fantasy? The answer to this question may require returning to a more fundamental level: data as a factor of production. China's National Data Administration has designated 2026 as the 'Year of Unleashing the Value of Data Elements.' On the soil of the industrial internet, the practices of pioneering enterprises have revealed a key logic – whether RAAS can truly succeed does not depend on the size of model parameters, but on an enterprise's ability to harness data as a factor of production.

We can analyze this logic through the following two typical RAAS cases, each in a different niche within the industrial internet: AI sales agents and SaaS procurement management.

Case One: Guangnian Chuda: AI Sales Agent Helps Companies 'Snag Orders' Globally

Founded in Shanghai in April 2025 by post-90s Tsinghua University graduate Pan Yiming, Guangnian Chuda has a clear and simple business: letting AI agents proactively find overseas clients for enterprises.

Its core product is the AI Sales Agent iSales. Unlike traditional SaaS tools, which companies purchase and then need to configure and staff themselves, often finding little actual value due to a lack of continuous operation, Guangnian Chuda delivers results, not tools. Clients don't need to hire additional personnel or send staff to learn system use and management. iSales automatically completes tasks like lead generation, email outreach, and intent qualification. The company only needs to follow up on the final qualified leads and pays based on actual results.

This model directly addresses the pain points of cross-border B2B enterprises in customer acquisition. Previously, most companies relied on offline trade shows or platform bidding, leading to a penetration rate of less than 20% for the previous generation of SaaS products in this market. Guangnian Chuda is different. Its customer acquisition service capability doesn't lie in sophisticated software but in its data capability. The backend of iSales integrates global commercial data sources, using AI algorithms to automatically analyze industry trends in overseas markets, the dynamics of relevant companies in various countries, and procurement signals. When the system identifies a clear procurement intent or business expansion plan from an overseas company, the Agent automatically generates multi-round personalized outreach strategies, completing the full-cycle loop from lead discovery to intent qualification at a very low cost.

With a clear and visible profit path, Guangnian Chuda completed a million-dollar angel round financing in March 2026.

Case Two: Vendr: Packaging 'Cost Savings' Itself as Result as a Service

If Guangnian Chuda represents an AI-native RAAS, then Vendr demonstrates a more traditional path: leveraging the compound interest of historical transaction data and expert wisdom to package 'cost savings' itself as Result as a Service.

Vendr is a US company headquartered in Boston, valued at $1 billion after its Series B funding in 2022. Its profit model is charging a management commission of 1% to 5% of a company's software spend. For a company with 500 employees, annual software procurement costs typically range from $2 million to $3.5 million. At a median rate of 2.5%, a single service engagement can generate $50,000 to $87,500 in revenue.

Client companies are willing to pay for its service because Vendr provides a cost-control guarantee: the platform ensures the software procurement savings for the client exceed its service fee. If not achieved, the service fee is fully refunded. This 'results-guaranteed' cooperation mechanism is essentially a mature reflection of RAAS in non-AI domains.

According to data disclosed on Vendr's website, the average client ROI is as high as 8x – meaning for every $100,000 in service fees paid, Vendr saves clients an average of $800,000.

Vendr does not extensively use the latest large language model technology. Its strongest barrier is its enterprise SaaS procurement price database. The company accumulates, day after day, real transaction price samples for various software categories under different combinations of client size, contract term, and add-on clauses. Vendr also deeply tracks the pricing strategy evolution of each major SaaS vendor – when they raise prices, under what conditions discounts can be obtained, what customer profiles are more likely to get favorable terms. These historical data spanning years form a deeply governed knowledge graph. When a new client requests service, Vendr's team and system, based on re-mining and result mapping of historical transaction data, output a precise cost-control plan for the client.

Because the price database is exceptionally robust and built on long-term governance experience, it is nearly impossible for any new competitor to build data assets of comparable depth in a short time. Furthermore, Vendr continuously tracks the actual procurement outcomes and negotiation practices of thousands of enterprises, forming an internal, continuously optimized standardized price analysis and effectiveness verification mechanism. The funds saved after each transaction are not only client gains but also become an iterative increment to Vendr's data assets.

02 The Core Logic of the RAAS Model: Dual Capabilities in Harnessing Data Elements

From these two cases, Guangnian Chuda and Vendr operate in different sectors – one AI-driven sales agents, the other traditional data-driven procurement services – yet they share a common foundation: the core competitiveness of their models stems from a systematic ability to harness data as a factor of production. This ability is primarily manifested in two mutually supportive dimensions: 'Application' and 'Governance.'

Generally, application capability refers to whether an enterprise can efficiently combine AI technology with data resources to deliver quantifiable business results at the product level. In the RAAS model, the direct manifestation of 'application capability' is a delivery loop that becomes smarter and more accurate with use.

Currently, China's AI application market is undergoing a historic shift from 'capability demonstration' to 'industrial application.' However, one of the core bottlenecks in the industrial internet sector is the insufficient 'model-data synergy' capability – meaning the deep integration of AI models with industry data resources is still in its early stages. When large model capabilities become highly homogenized, those who can more tightly and intelligently combine their own data assets with model capabilities will deliver more accurate and reliable results.

Guangnian Chuda's value precisely comes from its continuous and efficient processing of global commercial data sources and its ability to adjust in real-time based on each result's feedback – a typical 'application capability'-driven model.

Governance capability refers to whether an enterprise can continuously manage, clean, structure, and securely govern data resources, forming a scarce and hard-to-imitate data asset moat. Application capability solves the problem of 'running faster,' while governance capability solves the problem of 'others not catching up.'

Vendr's practice fully reveals the value of governance capability. Its core asset is not algorithmic models but the software procurement price data accumulated over years of operation. Under the dual influence of compliance protection and the scarcity of industry data, this data becomes a wall that cannot be easily crossed. Any new entrant trying to replicate Vendr's price database and negotiation experience from scratch would find it extremely difficult, even with massive investment of capital and time – because Vendr's data is the natural sedimentation of real transactions from thousands of clients over the past several years, possessing significant timeliness and exclusivity.

Similarly, Guangnian Chuda's long-accumulated overseas B2B customer acquisition data assets serve both as fuel for continuous product iteration and as a 'first-mover barrier' preventing copycats from easily replacing it with generic large models.

In fact, from the 'application capability' dimension, Guangnian Chuda's value does not lie in calling a 'larger' model, but in achieving deep integration between global commercial data resources and its sales agent. While most other AI sales projects directly call generic large models plus paid databases, Guangnian Chuda's founder, with an algorithm-driven mindset, abstracted the commercial process into algorithms for optimization, choosing a technical path with a higher implementation barrier but better results. This is precisely the embodiment of 'application capability' – through a continuous feedback data loop, every opened email, every lead converted into a valid opportunity becomes nourishment for training the model's next tasks, making the AI smarter and more accurate with use. Applying AI to the pain points is the manifestation of this capability.

Of course, application capability requires governance for support. From the 'governance capability' dimension, Guangnian Chuda's core barrier lies in the depth of governance of its accumulated industry and customer behavior data. For the highly vertical scenario of B2B goods trade, enterprises need to handle vast amounts of non-standardized information – trade policies of different countries, procurement cycles of different industries, decision-making habits of different enterprises. This data can only be transformed into a knowledge graph of sales strategies understandable by the Agent after long-term cleaning, labeling, and structured governance. It is nearly impossible for any new entrant to replicate data assets of comparable depth in this vertical field in a short time, because Guangnian Chuda's governance foundation was naturally sedimented and iterated through serving real clients. This is precisely the competitive moat built by 'data governance capability.'

In the field of AI applications, there is a repeatedly verified observation: the error rate of AI in enterprise core business scenarios (such as risk control, compliance, tender review) remains high – this is also the actual finding from recent interviews with multiple industrial internet enterprises by Ebrun Think Tank. The lack of a clear entity responsible for data governance and standard system construction is considered a major bottleneck. This conversely confirms that without a solid data governance foundation, the 'result delivery' promise of RAAS is difficult to realize.

From another perspective, we can see that the joint optimization of data governance and data application, in fact, constitutes the internal moat of the RAAS model.

03 Data as the Anchor: The Competitive Rules of Industrial Internet in the Present and Future

Although Sequoia Capital's RAAS narrative has sparked intense debate regarding its implementation pace and commercialization path, the direction it points to is undoubtedly correct: future business competition will comprehensively shift from 'selling tools' to 'delivering results.' But who will become the ultimate winners of this wave? The answer does not lie in the parameter race of models, but in the depth of an enterprise's mastery over data as a factor of production.

The strength of application capability determines whether your RAAS product can function effectively and be used well. The level of governance capability determines whether you can maintain market advantage and widen the gap with followers after achieving functionality.

RAAS lacking application capability is merely a soulless digital shell, unable to deliver the value clients seek. RAAS lacking governance capability is a gold mine without walls, which any competitor with sufficient capital can quickly approach within a relatively short time.

Data elements, through the four pathways of transforming innovation models, reconstructing industrial ecosystems, optimizing factor allocation, and catalyzing new business formats, are becoming the core engine driving the high-quality development of the real economy. On the soil of the industrial internet, this logic is being repeatedly validated by the real-money practices of enterprises like Guangnian Chuda and Vendr.

In the future, the winners may not necessarily be the providers of the largest general-purpose AI services, but more likely those who can establish data flywheels in vertical scenarios and continuously fortify data governance barriers – the controlling RAAS enterprises. They not only know how to leverage AI and data resources to stimulate application capability but also understand how to solidify this competitiveness through persistent governance building. The moat no longer stems merely from technical capability, but from deep binding with customer needs and the path of least resistance in usage.

In 2026, as AI accelerates its demystification, the cornerstone of this moat has become clearly visible – data elements. Anchored by data and sailing with result services, those enterprises wielding the dual-core capabilities of data can truly navigate the RAAS wave steadily and far.

Ebrun Think Tank will continue to focus on the development of the industrial internet and the enhancement of enterprise competitiveness in data elements, reporting on new achievements and cases in related developments. Contact email: huangbin@ebrun.com