480. Agentic AI vs. The SaaSpocalypse

Check out host Bidemi Ologunde's new show: The Work Ethic Podcast, available on Spotify and Apple Podcasts.

Email: bidemiologunde@gmail.com

In this episode, host Bidemi Ologunde breaks down the recent "SaaSpocalypse," the sudden market shock that sent software stocks tumbling as investors grappled with a new reality: what happens when AI companies like OpenAI and Anthropic start offering capabilities that look a lot like traditional SaaS, but at a fraction of the cost? Why did so many once-defensible software businesses lose huge chunks of value in just days? Are AI agents about to crush the per-seat software model, or is this panic getting ahead of the facts? And in a world where the software can now do the work, what exactly will customers still be willing to pay for? 

Sponsors and partners:

Promeed: 100% mulberry silk pillowcases and bedding that feel incredibly soft, stay breathable, and are naturally gentle on hair and skin.

SurviveX: professional-grade FSA/HSA eligible first aid and preparedness kits designed in Virginia, USA and produced in an FDA-registered facility.

Alison US CA: Alison is the world's largest free online learning and skills-training platform, helping more than 50 million learners in 193+ countries build career-ready skills with 6,000+ free courses, certificates, and diplomas.

eSign (iOS only): eSign is a clean, privacy-first document-signing app that works entirely on your device, letting you sign PDFs, DOCX files, images, and scans, edit and assemble pages, and export crisp 300 DPI PDFs in seconds, without accounts, cloud uploads, or compromising sensitive documents.

Transcript

SPEAKER_00 0:16

In early February 2026, investors abruptly repriced large swaths of the software market after Anthropic and OpenAI rolled out agenic features that don't just help users use software. They increasingly do the work that software was bought for. So the shorthand on trading desks and in analyst circles became SaaS Pocalypse, a fast fear-driven sell-off rooted in one idea. If AI agents can complete workflows end to end, the classic proceed SaaS model faces immediate pricing pressure and longer term uncertainty. What made this moment different from yet another AI headline was the convergence of three signals. First, role-based plugins and enterprise platforms from AI labs. Second, visible overlap with high margin SaaS categories such as research, legal, analytics, productivity, and support. And third, market moves large enough to force portfolio managers to react. The likely outcome is not SaaS dies, but SaaS changes. Pricing shifts away from pure seed counts, the UI shifts toward conversational or agent interfaces, and defensibility shifts toward proprietary data, compliance, and workflow ownership. So here's the kind of text message that can actually move markets. Tech reporter Dominic Madori Davis described a moment in which a founder messaged an investor. He was replacing his entire customer service team with clawed code, software that can write and deploy other software autonomously. The investor read it as something bigger than one founder cutting costs. It felt like a build versus buy switch flipping, where companies stop assuming a big SaaS contract is the default and start assuming they can build workflows with AI instead. So now let's zoom out from that text thread to the public markets. On Tuesday, February 3rd, 2026, investors reacted to new AI tooling, especially legal automation plugins, by dumping stocks tied to information heavy, workflow-heavy subscription software. One company that owns a flagship legal research product, Thompson Reuters, was down nearly 18% and headed for its biggest daily loss on record. The story of the SASPOCalypse is basically that gap between one founder's text message and a trillion dollar question. If agents can do the work, what exactly are we paying per user per month for? The SASPOCalypse didn't come out of nowhere. It landed after a year plus of increasingly capable agents marching up the stack from chat to browser control to full workflow execution. In December 2024, OpenAI introduced a$200 per month Chat GPT Pro tier and explicitly framed it as a place where more compute-intensive productivity features would arrive over time. On January 23, 2025, OpenAI launched Operator, an agent that can use its own browser, scrolling, clicking, and typing on web pages to complete tasks. On February 2nd, 2025, OpenAI launched Deep Research, pitched as an agentic capability that can do multi-step research and produce analyst-style reports after scanning large numbers of sources. On July 17, 2025, OpenAI unified these primitives into Chat GPT agents, explicitly describing an agent that can complete complex workflows end-to-end using its own virtual computer and even deliver editable slide decks and spreadsheets. By September 2025, the Build Your Own Impulse was already visible in the wild. Klarna was publicly rethinking its AI posture after earlier moves that included replacing tools such as Salesforce with AI build systems, an early, high-profile example of a company testing whether AI could substitute for major SaaS products. Then, late January 2026 poured gasoline on the debate. Anthropic rolled out co-work plugins designed to customize agents for specific roles and workflows. Think sales preparation, financial analysis, and legal review packaged so that teams can install them and let the agents run multi-step work. On February 3rd, 2026, the market reacted and SASPOCLIPS became a label for the sell-off. So two days later, on February 5, 2026, OpenAI announced Frontier, an enterprise platform to build, deploy, and manage AI agents, with early adopters named, and a strong emphasis on permissions, boundaries, and operationalization. Also on February 5, 2026, Anthropic announced Cloud Opus 4.6, explicitly highlighting stronger, sustained agency work and the ability to run analysis and create documents, spreadsheets, and presentations, including within co-work. By March 2026, Anthropic's release notes emphasized computer use inside co-work, opening files, navigating the screen, and acting on a user's behalf, meaning the agents can operate across tools even without a dedicated integration. So the key idea investors reprised was simple. The AI labs were not just selling intelligence, they were shipping the application behavior that used to justify whole SAS categories. Anthropic's co-work pitch is delegation to deliverables. Give it a go, and the agent works across your computer, local files, and applications to produce a finished deliverable, including tasks like organizing files, drafting structured reports, synthesizing research, and extracting data from dense documents. Co-work plugins take that one step further. The package, skills, connectors, and sub-agents, so the same underlying agent shows up as a specialist, legal, finance, or sales, without the buyer stitching tools together one prompt at a time. OpenAI's ChatGPT agent made the overlap even more explicit. It can navigate websites, run analysis, and produce editable spreadsheets and slide decks from a single instruction. Tasks that commonly span multiple SARS tools such as research, analytics, presentation, and workflow. And Frontier widened the aperture from individual productivity to enterprise operations. OpenAI described a platform integrated with systems of record with enterprise governance, explicit permissions, and auditable actions, and then framed it around deploying agents into business processes like revenue operations, customer support, and procurement. So this is why the February shock wasn't limited to app software. The market started asking whether the application layer itself, where SaaS companies traditionally monetize, was being invaded by model providers that could offer good enough versions of so many workflows, often bundled into one subscription. On February 3, 2026, the sell-off was immediate and specific. Thomson Reuters fell by nearly 18%, Relics fell about 14%, and Walters Cluer about 13%. In the same wave, firms adjacent to Knowledge Work in a Box got hit. Fact Set fell 10.5%, Morningstar lost 9%, and LegalZoom slumped 19.7%. The tremor spread beyond pure software into adjacent services and advertising names, including Omnicom and Publices, which Reuters noted also sold off amid AI threat fears. By February 4, 2026, Reuters quantified the broader drawdown, about$830 billion in market value erased since January 28, with the SP 500 Software and Services Index down nearly 13% over six sessions, and down 26% from an October peak, while investors debated whether the move was overdone or a true regime change. Other analysts described a similar magnitude with slightly different yardsticks. Forrester said that in the first week of February 2026, over$1 trillion of software market capitalization was erased in seven days, driven by the pace of AI agent innovation and fears that AI threatens to replace software workflows. And in the weeks that followed, the pain showed up in multiples and expectations, not just headlines. UBP pointed to valuation multiple compression as the dominant story, citing SAS EV and sales multiples around 5.8x versus a long-term average near 9.6x, framed as investors reassessing durability in a world where AI could commoditize parts of application software. The private market echo was quieter, but it was real. TechCrunch reported that venture-backed SaaS IPO filings were effectively absent and that some late-stage private SaaS companies were feeling pressure, with some struggling to raise extension rounds amid public market volatility and shifting expectations. Meanwhile, individual public SARS narratives became case studies. Asana, for example, was described as down over 50% in 2026 at the time of reporting, with its CEO positioning the firm as an orchestration layer for humans and AI agents and experimenting with hybrid pricing tied partly to AI usage and outcomes. The investor logic behind this pocalypse had three main legs seats, modes, and terminal value. First, proceed pricing implicitly assumes the workforce is human. If AI agents do more of the work, companies may need fewer seats, even if the underlying workflow still matters. TechCrunch framed it bluntly. When employees ask their AI of choice to pull data from systems, the proceed model starts to break down. Second, investors feared modes narrowing as AI lowers barriers to building software. According to TechCrunch reporting, customers now carry a negotiation weapon. If they don't like the price, they can build an alternative, pushing downward pressure into renewals, even if buyers don't fully replace a vendor. Third, the shock wasn't just about next quarter's ARR. It was about discounted cash flow map. Reuters described portfolio managers scrambling because rapid AI advances, muddy valuations, and business prospects beyond standard three to five year forecasts. Public company filings reinforced that boards and CFOs are internalizing the same risks. In its fiscal 2026 10K, Salesforce warned that attrition could increase due to factors including decreases in the number of users at our customers and the increased prevalence of consumption-based pricing models. And separately noted that pricing and packaging for AI offerings may not be widely accepted. In its fiscal 2025 10K, DocuSign explicitly cautioned that disruptive technologies like generative AI may alter demand, that AI could lower barriers to entry, and that the company might need to reduce or change its pricing model to remain competitive, describing pricing shifts tied to its intelligent agreement management platform. But there were counterpoints, some technical, some economic. One was data and trust still matter. Reuters noted skepticism that large language model vendors have the specialized data crucial in many verticals, a reason some analysts said the sell-off looked overdone. Another was that AI isn't replacing software, it's changing it. Reuters reported Nvidia CEO Yensin Huang calling it illogical to think AI would replace software and related tools. And closer to the app layer, Reuters cited JP Morgan software research head Mark Murphy, saying it feels like an illogical leap to assume one new plugin would replace every layer of mission-critical enterprise software. In other words, the market may have been right about pressure, but early about replacement. So now here is a practical question. What do we do with all this information on a Monday morning? First, recognize the pricing shift is already baked into risk disclosures. Salesforce's 10K points straight at consumption-based models and changing number of users dynamics. The implication is hybrid monetization, seat plus usage plus outcome, with AI features priced like capacity rather than headcount. Second, treat agenc platforms as a new distribution layer. Frontier is explicitly described as an enterprise platform integrated with systems of record, with identity, access management, and auditable actions built to run agents in production across business context. That is a map of where enterprise budgets are heading. Third, differentiate on governance and regulated workflow reality because agents still have sharp edges. Anthropic's co-work documentation notes that for team and enterprise plans, co-work activity is not captured in audit logs or compliance exports, and it explicitly warns against regulated workloads like HIPAA and FedRAMP. That gap is opportunity for SARS incumbents, be the system of record and governance layer that agents must operate through, permissioned, logged and auditable, even if the front end becomes conversational. Fourth, expect consolidation and flight to defensibility. Fortune argued that the public multiple compression is making take privates cheaper, while deal making remains strong, citing Enterprise SARS M ⁇ A value and framing the SASPOCalypse as an accelerant for consolidation. You can see this in real deals framed around AI era infrastructure. IBM announced an agreement to acquire Confluence for$31 per share in cash, valuing the company at$11 billion, explicitly tying the rationale to data for applications and AI agents. By March 17, 2026, IBM said it completed that acquisition, again framing it as a smart data platform giving AI models, agents, and automated workflows real-time trusted data. Similarly, Clearwater Analytics agreed to be taken private in a transaction valued around$8.4 billion led by Promira and Warburg Pinkos with$24.55 per share in cash. Fifth, big regulation and ethics into product strategy, not PR. Both Salesforce and DocuSign explicitly point to emerging AI legal frameworks, including the EU AI Act, as material compliance and business risks alongside broader privacy and IP concerns. And the IP side is not theoretical. A federal judge in Delaware ruled in February 2025 that a former competitor was not permitted to copy Thumbton Reuters' content to build a competing AI-based legal platform, an early landmark on fair use in AI-related copyright litigation. So the takeaway for SARS is this proprietary data rights, licensing, and defensible data sets are not nice to have. They are the leverage. So here is my takeaway for the SAS Pocalypse. It wasn't just a stock sell-off. It was Wall Street pricing in a new interface for work where you don't click through 10 software tabs. You simply delegate to an agent. And once you delegate to an agent, the unit of value changes. It's not how many people have logins, it is how much work got done and how safely can we let it run. So no, SaaS isn't dead. But bad SAS, the stuff that is easy to replicate, the stuff that is easy to swap, and priced like every user is a full-time worker. That's the part getting repriced. And the winners will look a lot like this. They own the trusted workflow, they hold the proprietary data, they offer compliance grade governance and they monetize outcomes instead of seats. Because in the age of AI co-workers, the software that survives won't just be the one with the best UI. It will be the one the agents are allowed to touch. If you like this episode, please share it with a relative, a friend, a coworker, a neighbor, an acquaintance, and so on. And then please leave a rating andor a review on your favorite podcast app. My name is PDM Logunde, and this is the Big Picture Podcast. Thank you for listening.