Ever since generative AI took off, there’s been a fierce debate: Will AI agents spell the end of traditional SaaS?
High-profile tech leaders like Microsoft’s Satya Nadella suggest that “agentic AI” – AI tools capable of automating and orchestrating tasks– could upend the SaaS model entirely. Skeptics on social media argue that there will be no reason to pay for software when AI can rebuild most SaaS tools and tailor workflows perfectly to specific use cases. Some even imagine a future in which software interfaces vanish entirely, with AI simply performing tasks on our behalf. Imagine an HR leader never manually processing payroll again or a sales manager never opening a CRM – an AI HR agent seamlessly calculates salaries, ensures compliance, and resolves discrepancies, while an AI sales agent logs interactions, drafts follow-ups, and forecasts revenue in real time.
We believe these narratives fundamentally mischaracterize why customers adopt SaaS tools in the first place. The foundational SaaS model is poised to endure in the coming AI age – albeit in an evolved form.
Best Arguments for SaaS Demise
There are compelling reasons to question SaaS’s resilience and to consider that we might have reached peak software.
First, there’s the potential for perfect customization. Externally built SaaS products, by nature, serve large numbers of customers rather than individual businesses. This disconnect has fueled disruption in the past, as seen when industry-specific, vertical software replaced horizontal incumbents. Salesforce, for example, is not right for all businesses; at Battery, we’ve successfully invested in vertical CRMs built specifically for home services, financial advisors and physical therapy, for example These platforms thrive by leveraging specialized data models, workflows, and industry integrations. If platforms like WebPT* offer more targeted solutions for physical therapy practices than generalized CRMs like Salesforce, it’s easy to envision AI pushing further – perhaps creating a hyper-personalized CRM uniquely designed for, say, a wound management specialist clinic in Albany, NY.
Another argument against traditional SaaS models revolves around the growing preference for simpler, conversational user interfaces over complex software workflows. There’s an old joke among Salesforce users that the software actually has two interfaces: the application itself, and the colleague you ask to check Salesforce for you. Users often prefer delegating tasks conversationally because it’s faster and more intuitive than navigating a cumbersome graphical interface. AI agents promise widespread, democratized access to conversational interfaces, potentially diminishing the value of detailed, pixel-perfect graphical UIs traditionally provided by SaaS providers.
Finally, the economics of SaaS adoption could shift dramatically as the cost of code approaches zero. Historically, businesses faced “buy versus build” decisions regarding software, with SaaS usually emerging as the more cost-effective choice because of economies of scale from shared development. But as AI automates coding tasks, development costs will plummet, eroding the traditional price advantage and margin structures that SaaS providers have relied upon. This commoditization could empower businesses to economically build customized software solutions, challenging the long-term value proposition of traditional SaaS offerings.
These arguments sound compelling, but we see things differently.
Why We Are Still Long SaaS
-
UI design, data models, and proprietary data still matter.
Consider the value of curation: Everyone has unique tastes in music, food, and fashion. Even if every apartment had a piano, people would still stream music on Spotify. Even with a fully stocked kitchen at home, getting a reservation at French Laundry or 4 Charles would remain difficult. Even if closet space were infinite, my wife would probably still not be happy with the clothes I pick. Infinite options do not replace expert-curated experiences. Similarly, even if code is unlimited and free, not every customer can effectively design software that delights users. When customers pay for SaaS, they pay for thoughtfully designed interfaces shaped by iterative user feedback and tasteful curation.
Picking the right data model is another subtle but significant advantage of SaaS. For example, Kustomer*, a customer service platform in our portfolio, stood out early by adopting a flexible NoSQL architecture that enabled rapid iteration and was built around a customer-centric data model, whereas incumbents were anchored to rigid, ticket-based relational structures that limited their adaptability and user experience.
Additionally, proprietary datasets represent strong competitive moats for SaaS providers. Companies like Gong* accumulate valuable, domain-specific insights over time, enabling richer experiences and superior predictive capabilities for its customers– advantages that are difficult for generic AI models that lack that specific context to replicate.
-
Not all workflows are best solved conversationally.
Although conversational UIs excel for certain tasks, we think graphical interfaces will remain essential for complex or information-rich workflows. Clicking buttons or visualizing data is frequently faster and cognitively easier than speaking or typing queries. Even as AI generates novel visualizations, customers typically prefer familiar, reliable experiences that minimize cognitive load.
-
Customer service and domain expertise
Companies purchase SaaS for more than just software – they invest in expertise and community. Customer support at Procurement Sciences*, a company whose technology helps companies manage the lifecycle of government procurement, is staffed by people who used to respond to government RFPs. HubSpot’s customer success team provides deep marketing and sales insights. Gong* representatives deliver specialized sales coaching knowledge. Customers pay for participation in these ecosystems and expert guidance as much as for the software itself.
-
Maintenance, updates, and ongoing costs
Maintenance and ongoing support are also crucial. Software built internally – AI-generated or not – still requires continuous updates, security patching, scalability management, and adaptation to changing regulations. SaaS providers offload these complexities from the customer, delivering stability and predictability. Companies like Veeva track evolving pharmaceutical regulatory requirements, while Wunderkind* ensures email compliance and deliverability. While AI could potentially automate some maintenance, customers highly value accountability from third-party SaaS providers (if nothing else, those vendors are entities you can hold to account when things go wrong!)
-
Industry standardization
Standardization through SaaS adoption is a feature, not a bug, and is a reason third-party SaaS tools remain popular. Familiarity with standard SaaS tools like Salesforce accelerates productivity for new hires. Particularly for non-differentiating functionalities like payroll or IT helpdesk systems, standardized SaaS reduces onboarding costs and boosts productivity.
It’s a Great Time to Build SaaS
Modern AI is undeniably transformational, but rather than signaling SaaS’s demise, it represents an opportunity for tremendous growth. Claims that AI will “kill SaaS” oversimplify how enterprises adopt and use technology. Even as internal builds become cheaper, customers aren’t merely purchasing code—they buy thoughtfully designed experiences, domain expertise, and reliable ecosystems.
That said, the pricing pressure posed by AI is real and will likely reshape SaaS monetization strategies. Pressured by AI, traditional per-seat pricing models may evolve toward consumption- or outcomes-based pricing, aligning provider incentives closely with customer outcomes. If you can deliver better outcomes than your competitors – via your proprietary data, partner ecosystem, or otherwise – you will retain pricing leverage, even as AI commoditizes aspects of software development.
The bigger picture is clear: AI isn’t shrinking the software market; it’s dramatically expanding it. As the cost of code approaches zero, software creation will explode. The long-running transition from manual workflows in spreadsheets to specialized software-defined workflows will only accelerate.
For SaaS CEOs and leaders, the mandate is clear: embrace AI aggressively, and double down on unique strengths like domain expertise, proprietary data, ecosystems, and curated user experiences. The challenge of building has gone from creating new features to curating product possibilities.
Ultimately, AI will be consumed much like software has been – as a service. Software continues to “eat the world,” as Marc Andreessen famously said, but AI is making our appetite even bigger.
The information contained here is based solely on the opinion of Brandon Gleklen, and nothing should be construed as investment advice. This material is provided for informational purposes, and it is not, and may not be relied on in any manner as legal, tax or investment advice or as an offer to sell or a solicitation of an offer to buy an interest in any fund or investment vehicle managed by Battery Ventures or any other Battery entity. The views expressed here are solely those of the author.
*Denotes a Battery portfolio company. For a full list of all investments and exits, click here.
The information above may contain projections or other forward-looking statements regarding future events or expectations. Predictions, opinions and other information discussed in this publication are subject to change continually and without notice of any kind and may no longer be true after the date indicated. Battery Ventures assumes no duty to and does not undertake to update forward-looking statements.