From Snowflake to Sierra, every enterprise software company is selling the same AI agent

The enterprise software industry is caught in an unprecedented melee, with traditional market boundaries completely shattered by artificial intelligence.

From database giant Snowflake to customer relationship management (CRM) leader Salesforce, nearly all major tech companies are racing to launch similar general AI agents, pushing former partners onto the stage of direct competition, creating huge growth opportunities while also presenting existential challenges.

The latest development in this trend is that companies with previously distinct business lines are now infiltrating each other's core markets. Salesforce, a company primarily focused on customer relationship software, recently launched an AI agent to address IT help desk issues. Meanwhile, IT service management software company ServiceNow has launched an AI agent aimed at sales personnel. Database provider Snowflake also released its AI agent product last month, claiming it can handle tasks across various professional roles, from sales to finance.

This "arms race" of AI agents has caused significant confusion among enterprise customers. With highly overlapping product functionalities, corporate buyers face an "extremely difficult" situation when making choices, leading some companies to delay large-scale purchasing decisions. This chaotic situation reflects the high stakes involved; as Snowflake CEO Sridhar Ramaswamy stated, software companies in this transformation will "either reach a trillion-dollar market value or go to zero."

This phenomenon also directly echoes recent market concerns about the "replacement of all software by AI." However, a closer look at market dynamics reveals that established software giants are building defensive fortifications based on their large customer bases and data accumulation, suggesting that the future industry landscape will be more complex than a simple "disruption-disrupted" narrative.

Market Overlap, Blurred Boundaries

The rise of AI agents is fundamentally rewriting the competitive landscape of enterprise software. At least seven major tech companies are currently engaged in direct competition across eight different functional areas, selling automated AI agents for roles in engineering, analytics, finance, marketing, sales, and customer service.

Part of the reason for this homogenized competition is that many startups and enterprise software giants rely on underlying AI models from companies like OpenAI and Anthropic to drive their agent products. Matt Luizzi, analytics director at Whoop, stated, "Every piece of software we use has launched its own AI agent solution," from Slack to Snowflake to Google Workspace, all promising to perform similar tasks such as data mining, sales forecasting, and even communicating with customers.

This trend means that established database and data streaming companies like Snowflake and Confluent are now competing with emerging AI application startups like Sierra and Decagon in areas such as sales and customer support agents "This is an exciting yet chaotic time," commented Jay Kreps, CEO of data management company Confluent.

Advantages of Software Giants: Data Convenience and Mixed AI Model Strategy

Faced with a flood of similar AI products, enterprise buyers generally feel at a loss. Ryan Teeples, Chief Strategy Officer of commercial accounting firm 1-800Accountant, stated that the current screening process has become exceptionally difficult due to significant overlap between tools, even though his company has started paying for some of Salesforce's agent tools.

In this chaos, existing software giants are leveraging their core advantage—data. They argue that customers find it more convenient to use agents that can directly extract data from their core software products (such as CRM applications or data warehouses). This strategy seems to be helping them gain an edge in winning initial trials from customers. Snowflake revealed last week that 1,000 customers are using its agent product Snowflake Intelligence, creating 12,000 agents.

Convenience has become key to enterprise decision-making. Pierre Matuchet, Senior Vice President of IT at European HR company Adecco, stated that their choice logic is very simple: "If the data is stored in Salesforce, we use Salesforce. If the data is outside of Salesforce, we consider other vendors." Similarly, Peter Stoltz, Chief Information Officer of tablet manufacturer reMarkable, said they chose Salesforce's Slack agent because most employees are already using Slack.

Although OpenAI CEO Sam Altman has suggested that AI agents may eventually completely replace work software like Slack, analyses from Wall Street and strategic layouts from industry giants paint a different picture. Goldman Sachs pointed out in a report that the current stage is similar to the software industry's transition from on-premises deployment to cloud computing, where AI is more likely to be a "force multiplier" for industry leaders rather than a pure disruptor.

To build a moat, enterprise software giants generally adopt a mixed AI model strategy. They combine domain-specific models trained on their proprietary data (such as Snowflake's Arctic model) with external cutting-edge large language models. This strategy locks customers into their familiar and deeply integrated ecosystems while maintaining flexibility.

Moreover, the "mission-critical" nature of enterprise software creates natural barriers. Goldman Sachs analysts noted that the "hallucinations" of AI models could lead to serious consequences in enterprise environments, making customers extremely cautious when migrating core business processes. This means that even if AI-native products are technically superior, it is challenging to achieve full customer adoption in the short term

Slow Adoption by Enterprises, Management and Collaboration of AI Agents Become the Focus

Despite the broad prospects, the commercialization of AI agents has not been smooth sailing. Currently, this new technology has not significantly boosted the revenue growth of companies like Salesforce, ServiceNow, and Microsoft. Salesforce CEO Marc Benioff recently downplayed previous claims about the ease of setting up agents, aligning with the industry consensus that enterprises are slow to adopt AI.

The reasons for slow adoption are multifaceted. First, configuring agents may require substantial human assistance, prompting companies like Amazon and Salesforce to invest additional personnel for support. Second, some companies believe that the AI provided by vendors is not mature enough. For example, Alex Devkar, Senior Vice President of Engineering at used car dealer Carvana, stated that they chose to develop their own AI chatbot because its performance surpasses that of off-the-shelf products.

In terms of business models, most vendors charge based on usage after a free trial period, typically costing between 20 to 30 cents per task. However, for companies like Snowflake, the current primary goal is to capture users rather than to achieve immediate profitability.

As the number of AI agents used internally by enterprises surges, a new challenge presents itself to all CIOs. “Everyone has to decide whether they need a centralized agent platform to coordinate all these agents scattered across different software,” said Eswar Veluri, Chief Technology Officer of fitness company Equinox. “We have not crossed that threshold yet.” This question suggests that the next round of competition in enterprise software may revolve around the management and collaboration of AI agents