The Disruption of Incumbent Enterprise Software Vendors is Imminent, Thanks to AI Agents.

Disrupting established software providers is a significant challenge for emerging startups. Customers face steep switching costs and are reluctant to entertain ripping and replacing due to extensive investments in implementation and customization over time.

Disruption can happen, however, as we saw when Salesforce disrupted Siebel Systems, the legacy on-prem CRM provider. As Santi Subotovsky of Emergence Capital writes, vulnerabilities emerge as companies expand, including bloated products and executives focused on providing shareholder value rather than solutions to customer pain points.

For true disruption of enterprise software incumbents, startups need to provide customers with technological innovation, new business models (that also tend to bring novel revenue opportunities), and a shift in distribution. Taking advantage of technological advancement is not enough to topple dominant players. When a young company can present all three, incumbents are unable and often unwilling to make the changes necessary to keep their customers, which is when disruption happens.

The shift from on-prem to the cloud illustrates this.

• Technological innovation: the onus of infrastructure shifted from the customer to the provider
• New business model: SaaS dramatically reduced opex for customers that no longer maintained the software internally and began paying monthly or annually per seat. Since software companies could support multi-tenancy and scale their product development efforts, they could offer new services and tiers like “freemium” much more efficiently.
• Distribution shift: from licensing to subscriptions

AI, the next generational technology shift, is upon us. Many startups are taking advantage of AI technological advancements to offer a new wave of features, including:

• Unstructured Data Utilization: Whereas traditional applications depend on structured data, the next wave of software will harness unstructured data as effortlessly as their predecessors handled structured inputs.
• Natural Language Interaction: Users can communicate their needs and queries to the software in everyday language, similar to how they would converse with colleagues, to pull insights from their data rather than relying on traditional reporting and dashboards.
• Customizability: Enterprise clients can leverage ready-made software modules to tailor applications to their unique needs without requiring deep technical knowledge or developer support.
• User-Friendly Design: Drawing inspiration from platforms like Airtable, Notion, and Slack, AI-native applications will be engineered to demystify complex processes for users, making technology more accessible.

Unfortunately, these features are not enough to disrupt enterprise software incumbents. Giants like Salesforce are the core repositories for critical company data, effectively locking in the organization, while employees become familiar with using these platforms, making them even stickier. Moreover, we have already seen incumbents integrate some of these features, taking advantage of their large datasets to deploy their own AI copilots to existing customer bases.

AI Agents: New business models and distribution channels

An AI agent can make decisions or perform actions automatically without human intervention. It observes its environment, makes decisions based on what it sees or knows, and then acts accordingly to achieve specific goals or tasks.

AI agents can apply the technology of generative AI while also bringing about new business models and distribution channels.

• Business models: The current model of per-seat pricing will no longer work, given the high processing costs of foundation models AI startups are built on. With Agents ostensibly replacing headcount, pricing will be closer to a fraction of a salary and compete with outsourced/international service providers and cover infrastructure. Given the technology gives them leverage to scale into new tasks faster, AI agent companies will also have the flexibility to offer more services at a lower cost, undercutting human capital-intensive providers. Here, there are significant people management and organizational implications to think about, which we plan to write about in the future.
• Distribution channels: Instead of software moving to the cloud, work shifts to the cloud through the advent of autonomous AI agents that can take actions and complete tasks independently.

This is only the beginning. Kojo Osei of Matrix Partners writes about AI Agents as a new distribution channel in which they represent buyers with purchasing power. In this world, “agent-driven commerce” would require businesses to optimize products for agents rather than humans, giving startups a leg up since they can incorporate this from the start.

While players like OpenAI, Google, Anthropic, and Cohere have dominated the development of foundation models, a clear disruptor has yet to emerge to shake up the enterprise software space by fully leveraging these models. Though still in its nascency, the companies poised to upend the current B2B software landscape will be those that go beyond simply integrating LLMs as supplementary tools. True disruption will come from pioneers who build their core products and services around advanced AI agents capable of autonomously performing complex workflows and decision-making traditionally handled by human teams. By pushing the boundaries of automation through these intelligent software systems, disruptors will gain an immense competitive edge over incumbent players still relying primarily on legacy platforms requiring extensive human involvement and oversight. The enterprises that embrace this AI-centric approach early on will be positioned to redefine how work gets done across industries.

Many questions remain about an AI-agent-driven world, including the potential impacts on the labor force. In the coming months, we will continue reflecting on these questions, and if you have an interesting perspective, we would love to hear from you!