Intuit has lost more than 40% of its market cap since the beginning of the year. It’s not alone. Many established SaaS players have seen their stock prices fall in recent months, including Adobe and IBM — the latter experiencing its most significant one-day drop (roughly $40 billion) with Anthropic’s announcement that Claude could now read, analyze and translate legacy COBOL into modern languages like Java and Python. The market has a name for it: the SaaSpocalypse.
The argument from investors and market watchers: AI agents can now do bookkeeping, file taxes and reconcile accounts — without a human ever touching software. For instance, instead of a human using QuickBooks to categorize transactions, Claude Cowork can access financial data, apply tax logic and autonomously prepare documents. Rather than using TurboTax, agentic AI tools can handle complex tax logic and even file taxes. In lieu of QuickBooks, automated agents can handle multi-step bookkeeping tasks (like lining up receipts).
Why investors are repricing SaaS
Intuit has been among the hardest-hit, with its market capitalization now sitting at around $106 billion.
The catalyst has been the emergence of fully agentic, no-code AI assistants like Claude Cowork and open-source tools like OpenClaw, whose founder was recently acqui-hired by OpenAI. Fears are that these cheaper service-as-a-service offerings (or service-as-software, or results-as-a-service, depending on who you ask) will upend pay-per-seat subscriptions; whereas traditional SaaS delivers a tool (software) for users to complete a task, service-as-a-service delivers a fully-automated outcome.
For instance, Anthropic’s Cowork platform includes finance capabilities that allow the agent to read financial files and turn them into structured models, tables and reports.
“The advantage is that I am abstracting away the complexity of my business operations,” said Brian Jackson, principal research director at Info-Tech Research Group (who prefers to call it “service-as-software”). “To hear about a model where you only pay when you get the outcome that you want, that’s very appealing.”
This emerging capability is in line with past technological advancements, he pointed out: IT departments used to be in charge of running infrastructure, but cloud computing came along to abstract away that management. Then, SaaS tools emerged to orchestrate the application layer. Now users manage their work — inputting data, filling out forms, creating analytics dashboards — within SaaS apps.
“So the next step is automated intelligence,” Jackson said. “Instead of having people do those things, we’ll just have AI do them.” Essentially, it could become a headless system without a UI; users simply let it run and don’t think about it.
This new concept comes at a time when enterprises are becoming fed up with the SaaS business model, he noted. Lock-in is frustrating, fees continue to go up, seats expand, and “it becomes this unwieldy operating cost,” Jackson said. “And it’s not always guaranteed to drive value, it doesn’t guarantee ROI at all.”
Why Intuit got hit the hardest
Intuit, which was founded in 1983, now serves around 100 million customers with a suite of products that, in addition to QuickBooks and TurboTax, include Mailchimp and Credit Karma. But these core offerings are now considered low-hanging fruit for AI, potentially endangering the company whose revenue model relies heavily on per-seat/per-user subscriptions.
Intuit’s CEO Sasan Goodarzi has recently shrugged off SaaSpocalypse claims, calling data the “most important moat” in a Semafor interview.
Marianna Tessel, EVP and GM for Intuit’s small business group, takes the same stance. Yes, Claude Cowork and similar agentic tools are “robust” tools, she noted, but Intuit has “persistent” and “durable” advantages.
Notably: First-party data. Customers generate various types of data on Intuit’s systems, whether it’s by creating an invoice, importing ledgers or performing various finance projects. Then there’s third-party data, which is generated through Intuit’s connections with 24,000-plus banks, e-commerce sites and other entities, Tessel pointed out.
AI agents simply do not have access to this “vastness” of data, she contended. Further, Intuit knows how to organize and use data, such as stitching together information across customer segments to provide market snapshots. “We understand this data, we know how to turn it into action,” Tessel argued.
She also doubled down on Intuit’s deep understanding of its customers. Rather than a chatbot that can process and act on numbers and figures, “we know what small businesses face,” she said, whether it’s their concerns around bookkeeping and payroll, or their struggles with hiring.
“We’ve been in business for over 40 years,” Tessel noted. “We have a lot of know-how that is very specific.”
Other SaaS companies stand staunchly behind this argument. Jon Aniano, Zendesk’s SVP of product and CRM applications, pointed out that his company serves 80,000 customers and deeply understands their needs. “We actually see [general purpose agentic tools] at a disadvantage because they’ve gotta go customer by customer and learn things that we’ve learned over the course of 20 years,” he said at a recent VentureBeat event.
The data moat argument does hold up, noted Info-Tech’s Jackson. He also pointed out that, realistically, the SaaS market is projected to grow at a “pretty good clip” in the years ahead. “Could that change very quickly? It’s possible, but it’s unlikely,” he said.
Also, SaaS is so entrenched in modern business, and pivoting to something entirely new can be a challenge. Even disruptive and compelling technologies like AI can take time to deploy at scale because enterprises have to recraft their workflows, Jackson noted.
“You have workers in place. You have departments in place. It just takes effort and time to change the processes and the expectations around these things,” he said, although “the appetite will definitely be there.”
How Intuit is betting on what agents can’t replicate
To get ahead of this, Intuit recently signed a multi-year partnership with Anthropic to bring AI agents to mid-market businesses. Using Anthropic’s Claude Agent SDK on the Intuit platform, enterprises will be able to build and customize agents. On the other end, Intuit’s tools can be surfaced directly inside Anthropic products such as Cowork, Claude for Enterprise, and Claude.ai through Model Context Protocol (MCP) integrations with TurboTax, Credit Karma, QuickBooks and Mailchimp.
This builds on Intuit’s previous rollout of Intuit Intelligence, which features specialized AI agents for sales, tax, payroll, accounting and project management. Users can query and interact with their financial data in natural language, automate tasks and generate dynamic reports or KPI scorecards.
“They have the data, they have the interface, and now they’re introducing themselves as an orchestration layer,” Jackson said of moves like this by large SaaS players. “We can be the place where you build your agents and manage them.”
To this point, Tessel calls Intuit “a well-run company” that can react with speed. Her team keeps up with orchestration advancements, reads academic papers and is “constantly learning” about new technologies. “We’re on it,” she said.”
Ultimately, companies must be “awake and aware right now,” she emphasized. As she put it: “What’s the pivot of the day? How many times did you pivot? Are you experimenting?”
Zendesk’s Aniano agreed that there are “cool new ways of developing software,” and acknowledged that he “lives” 90 to 120 minutes of his day inside Claude Code. Companies that can make the “mental shift” to building software in new ways can create a level playing field between incumbents and startups.
One thing that’ll be interesting to see is how quickly SaaS providers offer MCP plugins or build their own within their software suites, Jackson noted. “How good will these SaaS providers be at supporting AI interoperability?” he said. “And what ways will they try to create friction or make it harder for enterprises to abandon their interface?”
