Pperea.aiResearch
perea.ai Research · 1.0 · Public draft

Dead SaaS

The Silent Replacement of Human Teams at the Companies You Use — Q1 2024 to Q2 2026, Named and Sourced

AuthorDante Perea
Emaildante@perea.ai
PublishedMay 2026
Length6,558 words · 30 min read
AudienceCIOs and VPs of Operations buying SaaS in 2026; founders evaluating which SaaS vendors to depend on; investors modeling the SaaS-to-AI-margin transition; product leaders evaluating their own AI-replacement deployments against the operating record.
LicenseCC BY 4.0

#Foreword: Half the Apps You Use

Half the apps you use will quietly replace their support team, their onboarding team, and their content team with agents. You won't notice for months. That was the seed thesis for this paper. By Q2 2026, the seed has produced a named register: Salesforce[1][2], Cloudflare[3][4], IBM[5], Atlassian[6], Duolingo[7][8][9], Shopify[10][11], Workday[12][13], Freshworks[14], Intuit[15]. Together they account for over 10,000 customer-facing roles publicly eliminated or reframed between Q1 2024 and Q2 2026.

The companies you use look the same on the homepage[1]. The human you used to reach is gone[16]. The AI that replaced them resolves volume well — Salesforce reports 380,000 conversations through help.agentforce.com with 84% no-human resolution[17][18] — but escalates trust-laden cases poorly, the Klarna arc that is now an industry pattern documented in The Klarna AI Postmortem. Gartner's Feb-2026 prediction[19] that 50% of AI-attributed layoffs will rehire by 2027[19] reads less as forecast than as guidance based on the register below[20].

The buyer side: Sierra raised $950M at a $15.8B valuation in May 2026[21][22], serving 40% of the Fortune 50; Decagon raised $250M at a $4.5B valuation in January 2026[23], serving 100+ corporate clients. Anthropic captured approximately 40% of the enterprise LLM market by January 2026[24]. The agent vendors and the SaaS companies that bought their agents are now the same story, told from opposite ends of the same workflow.

#Executive Summary

The register documents five named patterns. First, the Cloudflare-pattern: explicit AI-replacement attribution[3][4]. CEO Matthew Prince and co-founder Michelle Zatlyn announced 1,100 layoffs on May 7 2026 attributed directly to AI doing the work — "Cloudflare's usage of AI has increased by more than 600% in the last three months alone"[4]. The stock fell 24% the same day despite beating Q1 2026 earnings[25].

Second, the Salesforce-pattern: consumer-facing customer support replaced[1]. CEO Marc Benioff said in September 2025 that Salesforce had cut its support headcount from 9,000 to 5,000[1][26] — 4,000 jobs eliminated as Agentforce handled the volume. Per Salesforce's FY25 Annual Report, Agentforce autonomously resolved 500,000+ support cases on the company's Help portal, with 84% no-human resolution[17][18]. A second wave of approximately 1,000 quiet layoffs in February 2026 hit the Agentforce team itself[27].

Third, the IBM-pattern: internal-tooling replacement[5][28]. The AskHR agent automated 94% of routine HR tasks; AskIT cut IT service interactions by 70%. 200 HR roles were replaced, but total IBM headcount grew because resources reallocated to programmers and sales[5].

Fourth, the Atlassian-pattern: self-fund-AI restructuring[6][29][30]. CEO Mike Cannon-Brookes cut 1,600 employees (10% of workforce) on March 11 2026 explicitly "to self-fund further investment in AI and enterprise sales"[6] — from a position of strength (cloud revenue +26% YoY, Rovo AI 5M+ MAU[16]).

Fifth, the Duolingo/Shopify-pattern: the cultural "AI-first" reframe[7][9][11]. CEO Luis von Ahn's April 2025 all-hands memo established five constructive constraints later echoed by Shopify CEO Tobi Lütke's "reflexive AI usage is now a baseline expectation" framing[10][11]. Headcount is gated on demonstrated automation impossibility.

The numbers Challenger, Gray & Christmas tracks[20] alongside Gartner's CEO survey data[19]: AI led job-cut reasons in March 2026 with 15,341 cuts, 25% of monthly total[20], and April 2026 with 21,490 cuts, 26% of total[31]. YTD 2026 AI-attributed cuts: 49,135 (16% of all cuts)[31][19], up from 54,836 in all of 2025 (5%)[32] and a cumulative 99,470 since tracking began in 2023[20].

The buyer-side bookend confirms the supply chain: Sierra at $15.8B valuation[21], Decagon at $4.5B[23], plus five named Anthropic enterprise deployments[33][34][35][36][24] together representing the vendor stack that powers the SaaS-workforce displacement.

#Part I: The Cloudflare Moment — Explicit AI-Replacement Attribution

On May 7 2026 at approximately 1 PM PT, Cloudflare CEO Matthew Prince and co-founder Michelle Zatlyn published a corporate blog post titled "Building For The Future"[4]. The post announced layoffs of more than 1,100 employees globally — approximately 20% of Cloudflare's then-5,156-person workforce[3]. The memo was sent verbatim to every employee, signed by both co-founders[4]. It was published one hour before the company's Q1 2026 earnings call[3][4].

The primary record of attribution is unambiguous. "We don't just build and sell AI tools and platforms. We are our own most demanding customer," Prince and Zatlyn wrote[4]. "Cloudflare's usage of AI has increased by more than 600% in the last three months alone. Employees across the company from engineering to HR to finance to marketing run thousands of AI agent sessions each day to get their work done. That means we have to be intentional in how we architect our company for the agentic AI era"[4]. The framing was not that AI would assist employees — it was that AI had made certain categories of employee unnecessary[3].

Prince was specific on the earnings call about which roles disappeared[25]. "We have seen that there are roles at Cloudflare that are not the roles we need for the future," he said[25]. "Just because you are fit does not mean you cannot get fitter. Over the last six months especially, the productivity gains from the people directly talking to customers and directly creating code have been incredible, and a lot of the support roles behind them are not going to be the roles that drive companies going forward"[25]. The language distinguished three groups: people who build the product, people who sell the product, and people who support the people who build and sell. The third group was eliminated[3].

The distinguishing claim against Meta's contemporaneous layoffs is precise. Meta's Mark Zuckerberg framed his cuts as capital reallocation: layoffs free up budget to buy GPUs[3]. Cloudflare's Prince framed them as AI doing the work: the GPUs are now the work[3]. The difference is between "we need your salary to buy GPUs" and "we no longer need you because the GPUs are doing your job"[3]. Cloudflare's announcement is the most explicit corporate attribution of layoffs to AI replacement on the public record as of May 2026.

The severance package signals the company's expected reputational cost[4][37]. Affected employees receive full base pay through the end of 2026, healthcare coverage through December 31 2025 in the United States, equity vesting extended to August 15 2026, and waived one-year vesting cliffs with pro-rated equity vesting through August[4]. Restructuring charges totaled $140-150 million: $105-110 million in cash severance and benefits, $35-40 million in non-cash equity-related expenses[3].

The market reaction was punishing despite strong fundamentals. Cloudflare beat Q1 2026 earnings estimates with 34% year-over-year revenue growth[3][4] and issued full-year guidance of $2.805-2.813 billion[3], beating consensus $2.79 billion[3][25]. The stock fell 24% the same day[3][37]. Prince told analysts on the earnings call that AI was "the biggest tailwind we've ever seen in Cloudflare's history" and the re-platforming of the internet around AI agents represents the company's largest growth opportunity[3]. "This is not about downsizing or saving costs," he said. "This is about having the right people in the right roles to build the future"[25]. The market disagreed.

Prince's closing prediction on the earnings call extended the claim beyond Cloudflare. "I would guess that in 2027 we will have more employees than we did at any point in 2026, but the roles are changing dramatically, and you have to do something dramatic to make that shift"[25]. If he is right, the 1,100 people losing their jobs at Cloudflare are early casualties of a restructuring that will reach every technology company[3]. If he is wrong, Cloudflare just fired 20% of its workforce during its best quarter, and the 24% stock drop is the market's way of saying so[3].

#Part II: The Salesforce Arc — Consumer-Facing Customer Support Replaced

On September 2 2025, Salesforce CEO Marc Benioff appeared on The Logan Bartlett Show podcast and delivered the most-quoted line of the SaaS-replacement era[1][26]. "I was able to rebalance my headcount on my support," Benioff said. "I've reduced it from 9,000 heads to about 5,000 because I need less heads"[1]. The 4,000-job reduction was attributed directly to Agentforce, Salesforce's autonomous AI-agent platform launched in the third quarter of fiscal 2025[38].

The mechanism is documented in Salesforce's FY25 Annual Report, filed with the SEC on March 5 2025[17]. "Internally, we are CustomerZero," Benioff wrote in the CEO letter. "Agentforce has already autonomously resolved over 500,000 support cases on our Help portal — 84% with no human intervention"[17]. The 84% no-human-intervention rate is the load-bearing operational claim — and the inverse, the 16% requiring human handling, is what the remaining 5,000 support agents handle. Sales productivity scaled in parallel: "speeding up sales quoting by 75%"[17].

Per Salesforce's 2025 Proxy Statement filed April 24 2025[18], the financial trajectory under Agentforce had matched the workforce reduction. "Data Cloud & AI annual recurring revenue reached $900 million, an increase of 120% year-over-year"[18]. "In the first 90 days after Agentforce launched, closed 5,000 Agentforce deals, including more than 3,000 paid"[18]. "Data Cloud surpassed 50 trillion records, which doubled Y/Y. Nearly half of the Fortune 100 are both AI & Data Cloud customers, and all of our top 10 wins in Q4 of fiscal 2025 included Data and AI"[18]. The headline FY26 revenue guidance was "$40.5 billion to $40.9 billion, up 7-8%"[17].

The Q1 2025 layoff cycle revealed the simultaneous-cut-and-hire pattern[39]. Per CRN's February 2025 reporting, Salesforce laid off 1,000+ employees (~1.5% of its 72,680 global workforce) WHILE hiring approximately 2,000 AI salespeople[39]. Benioff in December 2024: "We're adding another couple of thousand salespeople to help sell these [AI] products. We already had 9,000 referrals for the 2,000 positions"[39]. The pattern was not net headcount reduction; it was role substitution. Benioff: "Productivity is no longer tied to workforce growth. This is intelligent technology that can be scaled without limits. And Agentforce represents this next evolution of Salesforce"[39]. The hiring freeze on engineering followed[39]: "We're not going to hire any new engineers this year. We're seeing a 30% productivity increase in engineering"[18][39].

The February 2026 follow-up disclosed the second wave[27][2]. Per Inc.com and Business Insider's reporting, approximately 1,000 Salesforce employees were laid off "quietly" — without a public announcement — in early February 2026[27]. The departments hit: marketing, product management, data analytics, AND the Agentforce AI division itself[27][2]. Adam Evans, former head of Agentforce, departed[27]. Five senior leaders exited between December 2025 and February 2026; six executives were promoted to fill the gaps[27][2]. The inclusion of the Agentforce team in the layoff round was particularly notable given Benioff's repeated framing of Agentforce as "the core of every product we make now"[27].

The cumulative impact: per Salesforce's FY25 Annual Report[17], 380,000 conversations through help.agentforce.com with only 2% requiring human escalation. AI agents internally doing 30-50% of all work at Salesforce, with 93% accuracy[18]. The mechanism by which 4,000 customer-support jobs were eliminated is documented in the SEC filings as efficiently as any cost-reduction case study on the public record[2][17][40][38]. The Risk Factor language in the FY25 10-K formalized the trade-off: "Social and ethical issues, including the use or capabilities of AI in our offerings"[2] — the first time Salesforce's annual report acknowledged AI ethical risk as a material operating risk.

#Part III: The IBM AskHR Pattern — Internal-Tooling Replacement

In May 2025, IBM CEO Arvind Krishna told The Wall Street Journal that the company had replaced approximately 200 human resources workers with AI agents[5]. The replacement was operationalized through AskHR, IBM's internal HR-automation agent[28]. Per IBM's own data, AskHR automated 94% of routine HR tasks including vacation requests, pay statements, and payroll questions[28]. A parallel system, AskIT, reduced internal IT service interactions by 70%[5].

The pattern is the internal-tooling variant of AI replacement[5]. Unlike Salesforce or Cloudflare, IBM's announcement did not announce a workforce reduction. Instead, Krishna framed the reallocation as net-positive: "Our total employment has actually gone up, because what [AI] does is it gives you more investment to put into other areas"[5]. Those other areas, per Krishna, included programmers and salespeople — roles requiring "critical thinking" where people need to "face up or against other humans, as opposed to just doing rote process work"[5].

The 2023 precedent contextualizes the May 2025 announcement[28]. In a Bloomberg News interview, Krishna projected that approximately 7,800 jobs — roughly 30% of IBM's then 26,000-employee HR/back-office workforce — could be replaced by AI and automation over a five-year horizon[28]. The May 2025 confirmation of 200 actual replacements is the first concrete data point on that trajectory.

The June 2025 follow-up disclosed 8,000 IBM layoffs (~3% of the company's then 270,300 global workforce per the 2024 annual report)[28]. While IBM framed these as part of broader workforce restructuring rather than direct AI replacement, the timing and language tied the moves to the same automation thesis. IBM CHRO Nickle LaMoreaux: "Very few roles will be completely replaced. Instead, AI will take over the repetitive parts of the job, freeing up employees to focus on areas that need human judgment and decision-making"[28].

The distinguishing pattern from Salesforce is consequential. Salesforce frames AI replacement as net-negative headcount reduction[1]. IBM frames it as net-positive headcount realignment[5]. Both are correct; the difference is who counts as "labor" in the equation[5]. At Salesforce, the 4,000 eliminated customer-support agents were on the payroll; at IBM, the 200 eliminated HR workers were also on the payroll, but the reallocation toward programmers and salespeople maintained total employment.

The IBM pattern generalizes to every enterprise with substantial internal-services workforces — HR, IT, finance, legal, procurement. By Q1 2026, the playbook had spread: Goldman Sachs deployed Anthropic-built agents for accounting and onboarding[41]; JPMorgan's Jamie Dimon told investors the bank has "huge redeployment plans" for staff displaced by AI[42], with operations staff down 4% and support down 2% while client-facing roles grew 4%. The IBM pattern's distinguishing claim is that the workforce shift is invisible to consumers and reframed internally as productivity, not displacement.

#Part IV: The Atlassian Self-Fund Pattern — Restructuring to Reallocate to AI

On March 11 2026, Atlassian co-CEO Mike Cannon-Brookes published a blog post announcing 1,600 layoffs[6] — approximately 10% of the company's then 16,000-employee workforce, leaving a post-cut total of ~14,400. The stated rationale was specific and unembarrassed. "We are doing this to self-fund further investment in AI and enterprise sales, while strengthening our financial profile"[6].

Cannon-Brookes's framing distinguished Atlassian from the Cloudflare-pattern[6]. "We fundamentally believe people and AI create the best outcomes. Our approach is not 'AI replaces people.' But it would be disingenuous to pretend AI doesn't change the mix of skills we need or the number of roles required in certain areas. It does"[6]. The acknowledgment that AI changes the role mix was treated as a defensible explanation rather than a defensible deflection — a public, on-the-record statement that the workforce shape and the AI investment thesis were coupled.

The financial backdrop was strong[16]. Atlassian reported Q4 2025 cloud revenue of approximately $1.067 billion (+26% YoY), remaining performance obligations of $3.814 billion (+44%)[16], 600+ customers generating over $1 million in ARR[16], and 5+ million monthly active users on Rovo, the company's AI productivity assistant[16]. The cuts came from a position of strength, not weakness — a fact that distinguishes the Atlassian-pattern from cost-cutting layoffs at companies under earnings pressure.

The restructuring charges totaled $225-236 million per the SEC filing[30]: $169-174 million in future cash outlays for severance, notice period, and benefits[30]; $56-62 million in office space exit charges[30]. The severance package was unusually generous: minimum 16-week separation + 1 week per year of service + $1,000 USD technology payment + 6 months extended healthcare + outplacement services[6]. The package size signals the workforce reduction is meaningful, not marginal, and that Atlassian budgeted for the reputational cost of the cuts.

The impact distribution per Reuters[30]: 40% of impacted employees in North America, 30% in Australia, 16% in India, with smaller cuts across Europe, the Middle East, Africa, Japan, and the Philippines. CTO Rajeev Rajan departed March 31 2026; co-CEO Cannon-Brookes signaled additional leadership transition around AI priorities[16].

The investor context made the timing legible[30]. Atlassian shares were down approximately 33% in 2025[30]. Wall Street investors had become explicit about questioning the future of seat-based enterprise SaaS as Anthropic's Claude Cowork and OpenAI's similar offerings began rolling out vertical plugins that compete directly with Atlassian, Salesforce, ServiceNow, and Workday[43]. The Atlassian-pattern was the financially disciplined response: prove AI-driven margin expansion now, before investors price the SaaS-disruption thesis into the stock further. The market's verdict was modestly positive — Atlassian shares rose 2% in extended trading the day of the announcement[30] — but the broader sector remained pressured.

#Part V: The Duolingo / Shopify "AI-First" Memo Pattern — Cultural Reframe

On April 28 2025, Duolingo CEO and co-founder Luis von Ahn published an all-hands email on the company's official LinkedIn channel under the header "Duolingo is going to be AI-first"[9]. The post received over 1,091 comments within days; the consumer reaction was overwhelmingly negative, with multiple commenters announcing app deletions and the end of multi-year streaks[9]. The memo established a template that Shopify, Box, and at least four other SaaS companies subsequently echoed in 2025-2026.

The five "constructive constraints" are quoted verbatim from von Ahn's memo[44]: (1) "We'll gradually stop using contractors to do work that AI can handle"; (2) "AI use will be part of what we look for in hiring"; (3) "AI use will be part of what we evaluate in performance reviews"; (4) "Headcount will only be given if a team cannot automate more of their work"; (5) "Most functions will have specific initiatives to fundamentally change how they work"[44]. The constraints operationalize the reframe at the personnel-management layer.

The mobile-first analogy was central to the memo's defense[9][7]. "Just like how betting on mobile in 2012 made all the difference, we're making a similar call now," von Ahn wrote[9]. "This time the platform shift is AI"[7]. Critics noted the analogy is contestable: mobile-first was about WHERE work happens (a product decision); AI-first is about WHO does the work (a personnel decision)[44]. The category mistake was widely flagged in the LinkedIn comments.

The Shopify parallel arrived three weeks earlier[11]. On April 7 2025, Shopify CEO Tobi Lütke posted an internal memo on X — initially leaked, then confirmed authentic by Lütke — with the operative line: "Before asking for more headcount and resources, teams must demonstrate why they cannot get what they want done using AI"[11]. Lütke's framing: "What would this area look like if autonomous AI agents were already part of the team? This question can lead to really fun discussions and projects"[11]. The Lütke memo and the von Ahn memo were widely paired in the trade press as the cultural-template moment for the SaaS sector.

The January 2024 Duolingo precedent contextualizes the April 2025 memo[8]. Duolingo had already offboarded approximately 10% of its contractor workforce in late 2023, primarily translators whose work was now being handled by GPT-4 with surviving human "content curators" reviewing outputs[45][46]. A Reddit thread from a Duolingo contractor described the team being cut from four members to two[45]. The April 2025 memo formalized what had begun in January 2024 as a one-off contractor reduction.

The productivity payoff was disclosed two days after the memo[47]. On April 30 2025, Duolingo launched 148 new language courses simultaneously — Spanish, French, German, Italian, Japanese, Korean, and Mandarin, each available in 28 different user interface languages[47]. Von Ahn: "Developing our first 100 courses took about 12 years, and now, in about a year, we're able to create and launch nearly 150"[47]. The 14x productivity gain in content production is the strongest argument for the AI-first reframe; the customer-side response (app deletions, broken streaks) is the strongest argument against it.

The structural pattern of the Duolingo/Shopify cultural reframe is reproducible. Per the LinkedIn news commentary[9], "First Shopify. Now Duolingo goes AI. Their CEO said they will become an 'AI-first' company. This means: AI usage is part of the hiring process. AI usage is part of performance reviews. Teams will only get headcount once they prove they can't automate more of their work. If you're a digital native business and haven't gotten the memo, here's the memo." The category lesson: the cultural reframe pattern works when the contractor model was already the dominant cost structure (Duolingo's case). It works less well when full-time employees must be terminated to match the reframe (the Klarna case documented in Paper #7).

#Part VI: The Buyer Side — Sierra, Decagon, and the $20B Agent-Vendor Stack

On May 4 2026[21], Sierra Technologies — founded in 2024[22] by Bret Taylor (former co-CEO of Salesforce and current chairman of OpenAI) and Clay Bavor (former Google VP of AR/VR) — closed a $950 million Series E at a $15.8 billion post-money valuation[21][22]. The round was led by Tiger Global and GV (Alphabet's venture arm) with participation from Benchmark, Sequoia, and Greenoaks[21]. The raise tripled Sierra's September 2025 valuation of $4.5 billion within eight months[22].

Sierra's growth trajectory is the financial mirror of the workforce displacement happening at its customers[22]. The company reported $100 million in annual recurring revenue by late November 2025 and $150 million by early February 2026[22] — "$150M ARR in 8 quarters, a growth timeline that is unprecedented in traditional software"[21]. The customer roster as of May 2026 includes 40%+ of the Fortune 50[21][22]: Prudential, Cigna, Blue Cross Blue Shield, Rocket Mortgage, "1 in 3 of the world's largest banks," plus consumer brands Sonos, SiriusXM, WeightWatchers, and Casper[21][48]. Sierra's agents are documented to handle "billions of interactions, from refinancing mortgages to processing insurance claims, managing returns, and powering nonprofit fundraising campaigns"[22].

Sierra's April 2026 product launch made the recursion explicit. Ghostwriter is "agent-as-a-service" — users describe what they need in natural language, and Ghostwriter autonomously creates and deploys a specialized agent to handle it[22]. The thesis Taylor articulated at HumanX in March 2026: "the end of the era of clicking buttons"[22]. Most enterprise software, in this framing, will be replaced by agents talking to other agents — and the human roles that interfaced with the old software become casualties.

Decagon — co-founded in 2023 by Jesse Zhang and Ashwin Sreenivas — closed a $250 million Series D at a $4.5 billion valuation on January 28 2026[23], up 3x from the $1.5 billion June 2025 round. The round was led by Coatue Management and Index Ventures with participation from Andreessen Horowitz, Accel, Bain Capital Ventures, Chemistry VC, Definition Capital, Forerunner, Ribbit, and Starwood Capital Group[23]. A subsequent March 2026 tender offer allowed 300+ employees to sell vested shares at the same $4.5 billion valuation[23][23].

Decagon's named customer profile per the CorePiper comparison and Bloomberg's coverage[23][48]: Duolingo, Notion, Rippling, Coda, Webflow, Avis Budget Group, Deutsche Telekom, 1-800-Flowers, Quince, Oura Health, and Away Travel[23][23][48]. The common thread per CorePiper: "high ticket volume, a well-documented product knowledge base, and a team that wants to reduce human agent headcount rather than augment it"[48]. Decagon's positioning is technical-SaaS-first; Sierra's is consumer-brand-first[48].

The buyer-vendor cross-pollination is what makes the pattern legible[48]. Duolingo is publicly an AI-first company (Part V) AND publicly a Decagon customer[48]. Sonos is publicly a Sierra customer AND has documented AI-driven CS workforce changes[48]. The vendors selling the agents and the SaaS companies disclosing the layoffs are now the same story told from opposite ends of the same workflow. Per Decagon's investor materials, "Gartner estimates there are 17 million contact center agents worldwide"[23] — Sierra and Decagon are competing to automate that global labor pool.

#Part VII: The Anthropic Enterprise Stack — Five Named Deployments

Beyond the customer-service vendors, the broader-stack enterprise AI deployment register is anchored by five named Anthropic customers between Q4 2025 and Q1 2026[33][34][35][36][24]. Each documents a different version of the AI-replaces-or-amplifies tradeoff[34][35][24].

Block (parent of Cash App, Square, Afterpay) uses Claude on the Databricks Data Intelligence Platform as the default model powering codename goose, Block's open-source agent[34]. As of the customer story publication, 4,000 of Block's 10,000 employees actively use goose, with adoption across 15 distinct job profiles[34]. Per Block's case data: 75% of engineers save 8 to 10+ hours per week using goose; adoption doubled in a single month with weekly user engagement growing 40-50%; Claude 3.5 Sonnet was the only model to consistently achieve 100% success on Block's benchmark tests[34]. Block's stated goal: 30% employee-time savings through AI in 2025[34].

Rakuten (Japanese e-commerce conglomerate with 70+ businesses) deployed Claude Code across its engineering organization with measured results: 79% time-to-market reduction (24 days → 5 days), 99.9% accuracy on complex code modifications, and a documented 7-hour sustained autonomous coding session on an open-source refactoring project[35]. Claude Managed Agents were deployed across product, sales, marketing, and finance within one week each[35]. Yusuke Kaji, General Manager of AI for Business: "It's about multiplying what each team can achieve, not just automating existing tasks"[35].

ServiceNow announced Claude as the default model for ServiceNow Build Agent and as a preferred model across the ServiceNow AI Platform[33]. The rollout extends Claude + Claude Code "across its global workforce of more than 29,000 employees"[33]. ServiceNow processes 80 billion workflows per year on its platform[33]. The deployment cut seller preparation time by 95% and targets a 50% reduction in time-to-implement for customers[33]. Build Agent usage is expected to quadruple over the next 12 months per ServiceNow's projection[33].

HUB International (insurance brokerage, 20,000+ employees) completed a full Claude Enterprise deployment in Q4 2025 — one of the fastest enterprise AI rollouts in financial services[36]. Early results per the PRNewswire announcement: 85% productivity increase in targeted use cases, 2.5 hours saved per employee per week, 90%+ user satisfaction[36]. HUB selected Anthropic specifically "for its state-of-the-art performance in complex reasoning and coding, the lowest hallucination rates in the industry, and robust security features purpose-built for highly regulated environments"[36].

Travelers Insurance (NYSE: TRV, 30,000+ employees) deployed the Claude 4 suite across its entire global workforce in Q1 2026 — what TokenRing AI's analysis described as "one of the largest enterprise-wide integrations of generative AI in the financial services sector to date"[24]. The $1.5 billion strategic AI bet anchors on two tiers: Claude Code for ~10,000 engineers, data scientists, and analysts; TravAI internal ecosystem for the broader workforce[24]. An automated email classification system on Amazon Bedrock now categorizes millions of customer inquiries with 91% accuracy[24].

The vendor-side market structure makes the pattern legible[24][36]. Per Travelers/TokenRing reporting[24]: Anthropic captured approximately 40% of the enterprise LLM market by January 2026[24][36], with a 50% share in AI coding[24][36]. Anthropic generates roughly 85% of its revenue from B2B contracts[24], distinguishing it from OpenAI's more consumer-weighted mix[24][34]. The Anthropic enterprise deployment register thus represents the broader-stack analog to Sierra/Decagon: the agents that don't appear on customer-service screens, but that quietly displace internal knowledge work across HR, finance, engineering, and operations at scale.

#Part VIII: Detection Signals — Eight Patterns for Grading Your SaaS Dependence

The named register from Parts I-VII condenses into eight detection patterns. A CIO buying SaaS in 2026 can grade vendor dependence and price switching cost using these signals.

Pattern 1 — Customer-service spend grew while headcount fell. Klarna documented this load-bearing inconsistency in Q3 2025 (Paper #7); Salesforce's broader operating expenses tell the same story. The headline number ("4,000 jobs eliminated, 853-FTE-equivalent AI productivity") obscures the unit-economics reality: per-transaction cost falls while aggregate spend rises with transaction volume growth[1][18]. Read line items, not press releases.

Pattern 2 — "Quiet" second-wave layoff cycles 12-18 months after the headline announcement. Salesforce's February 2026 ~1,000-employee cut followed the September 2025 4,000-cut by approximately five months[27][2]; the second wave hit the Agentforce AI team itself, signaling that the first-wave architectural choices were being revised internally before the company would publicly confirm. Expect the second wave; budget for the timing.

Pattern 3 — Customer support spend disclosed as a line item that doesn't shrink. Klarna's $42M → $50M Q3 2025 disclosure is the canonical example: CS spend grew 19% YoY despite the AI doing 22% more work. The pattern recurs at Salesforce where AI handles 30-50% of internal work at 93% accuracy, yet the company simultaneously hired 2,000+ AI salespeople[39][18]. The line-item reality contradicts the headline replacement narrative more often than press releases suggest.

Pattern 4 — Internal AI usage growth percentages quoted in press releases. Cloudflare's "600% in three months" is the template[4]. Watch for the metric companies don't usually report (internal AI session count, engineer copilot usage rates) suddenly becoming the headline. Per Salesforce's FY25 Annual Report, the company disclosed that internal "AI agents are now doing 30-50% of all work within Salesforce itself" with 93% accuracy[18][17]. When the metric appears, the workforce announcement is downstream by 6-12 months.

Pattern 5 — Verb shift across CEO earnings-call language: "replace" → "augment" → "complement" → "amplify." The pivot signals architectural reframe rather than continued cost reduction. Klarna walked this arc February 2024 → February 2026 (Paper #7); Atlassian's Cannon-Brookes explicitly disowned "AI replaces people" framing while cutting 10% of workforce[6]; Cloudflare's Prince refused the augmentation framing and stayed on "support roles are not going to be the roles that drive companies forward"[25]. The vocabulary the CEO uses on a given quarter's earnings call telegraphs the next quarter's workforce decisions.

Pattern 6 — Severance package size signals expected headcount reduction is material, not marginal. Cloudflare's full base pay through end of 2026 + healthcare through December 31 + equity vesting extended to August 15 + waived one-year vesting cliffs[4] is what mass layoffs look like at companies that fear PR backlash AND need to retain remaining talent. Atlassian's 16-week-minimum + 1-week-per-year-of-service + $1,000 USD technology payment + 6 months extended healthcare[6] is the same signal at smaller scale. The package size is a leading indicator of the cut's strategic importance.

Pattern 7 — The "AI-first" cultural memo precedes contractor/role elimination by 6-12 months. Duolingo's April 28 2025 memo[9] preceded the broader contractor phase-out arc[7]; Shopify's April 7 2025 Lütke memo preceded multiple operations-team reorganizations in 2026[10][11]. If your SaaS vendor's CEO publishes an AI-first memo, model the workforce impact 6-12 months out and price your switching cost accordingly.

Pattern 8 — The buyer-vendor cross-pollination signal. Is your SaaS vendor a publicly-named Sierra, Decagon, or Anthropic enterprise customer? If yes, model the timing carefully: vendor deploys agent platform → 6-12 months later, vendor announces "we don't need to backfill these roles." Duolingo became a Decagon customer before announcing its AI-first restructuring[48][9]; the pattern is documented and predictable[21][22][23][34][35][36].

The CIO buyer playbook (board-ready, in eight steps): (1) Demand vendor's Sierra/Decagon/Anthropic deployment status as a disclosure item before renewal. (2) Score AI-replacement risk by line-item analysis of customer-support and operations spend YoY. (3) Track verb-shift in vendor earnings calls. (4) Negotiate exit clauses tied to disclosed AI-restructuring milestones (the Klarna pattern of "lower quality" admissions). (5) Build dual-vendor resilience for any vendor that has published an AI-first memo[9][11]. (6) Model customer-facing degradation at 12 and 18 months out (the Klarna 14-month arc as the reference timeline). (7) Maintain an internal "AI-replacement-stage" score per vendor (Stage 1 announcement → Stage 2 first cycle → Stage 3 quality reversal → Stage 4 hybrid reframe). (8) Treat AI-driven cost-savings vendor claims with the same skepticism as the headline Klarna 700-FTE claim was disproven by the Q3 2025 $50M CS-spend line item.

Per Gartner's 350-CEO survey of $1B+ companies deploying AI agents[31]: "Workforce reduction may create budget room, but it doesn't create return. Organizations that improve ROI are not those that eliminate the need for people, but those that amplify them." The companies winning with AI are investing in people alongside the technology, not instead of them. The companies whose SaaS you buy in 2026 will increasingly look like the eight-pattern register above — or they will be replaced by the ones that did.

#Glossary

AI-first memo. A CEO all-hands message establishing that AI usage is required across hiring, performance reviews, and headcount approval. Coined by Duolingo's Luis von Ahn (April 28 2025); echoed by Shopify's Tobi Lütke (April 7 2025); applied by Box, Klarna, and at least four other SaaS vendors in 2025-2026.

Reflexive AI usage. Lütke's April 2025 phrase: "reflexive AI usage is now a baseline expectation." The framing requires employees to demonstrate that work cannot be done with AI before headcount is approved.

Agentic AI-first operating model. Cloudflare's May 7 2026 phrase (Prince + Zatlyn blog post) describing the architectural reorganization that justified the 1,100-job cut. The framing inverts traditional org structure: AI agents are first-class organizational citizens, humans are deployed where AI cannot perform.

Agentforce arc. Salesforce's September 2024 - February 2026 sequence[18]: launch (October 2024) → 5,000 deals in 90 days[18] → 4,000 customer-support layoffs (Sept 2025)[1] → 380K conversations 84% no-human[17] → 1,000 quiet Feb 2026 layoff hitting the Agentforce team itself[27].

AskHR pattern. IBM's internal-tooling AI replacement: 200 HR roles eliminated via AskHR agent (94% routine task automation) while total headcount grew through reallocation to programmers and salespeople.

Build Agent (ServiceNow). ServiceNow's autonomous-agent builder running Claude as default; rolling out across 29,000+ ServiceNow employees Q1 2026.

Codename goose (Block). Block's open-source AI agent powered by Claude on Databricks. 4,000 of 10,000 Block employees use it; 75% of engineers save 8-10+ hours per week.

Managed agents (Anthropic). Anthropic's enterprise-grade agent platform: deployed at Rakuten across product/sales/marketing/finance within one week per function.

Ghostwriter (Sierra). Sierra's April 2026 "agent-as-a-service" product: users describe what they need in natural language; Ghostwriter autonomously creates and deploys the specialized agent.

Quiet AI layoff. Workforce reduction announced without a public statement, surfaced through LinkedIn employee posts. Salesforce February 2026 ~1,000-employee cut is the canonical example.

AI cost-cut vs AI capital-reallocation. The Cloudflare/Meta distinction. Cloudflare: AI does the work, so the budget AND the work are reallocated. Meta: AI requires GPU capital, so salaries are reallocated but the work persists.

Verb-shift detection. Tracking CEO earnings-call language migration from "replace" → "augment" → "complement" → "amplify" as a leading indicator of strategic reframe.

AI-replacement-stage score. Four-stage maturity ranking per vendor: Stage 1 announcement; Stage 2 first cycle of cuts; Stage 3 quality reversal (the Klarna inflection); Stage 4 hybrid layered architecture (the Klarna 2026 equilibrium).

#References

References

  1. Emma Roth / The Verge (2025-09-03), Salesforce cut 4,000 customer support roles after deploying AI agents. https://www.theverge.com/news/770068/salesforce-cut-4000-customer-support-roles-after-deploying-ai-agents ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  2. Salesforce, Inc. (2025-03-05), FY25 10-K Annual Report. https://s205.q4cdn.com/626266368/files/doc_financials/2025/q4/3d22b243-78ec-4635-9b99-a0a33967944d.pdf ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  3. Cristian Dina / TheNextWeb (2026-05-08), Cloudflare cuts 1,100 jobs in agentic AI pivot despite beating Q1 2026 earnings as stock falls 24%. https://thenextweb.com/news/cloudflare-layoffs-agentic-ai-earnings-stock ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13 ↩14 ↩15 ↩16 ↩17

  4. Matthew Prince + Michelle Zatlyn / Cloudflare (2026-05-07), Building For The Future. https://blog.cloudflare.com/building-for-the-future/ ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13

  5. ETHRWorld via Wall Street Journal (2025-05-10), IBM replaces 200 HR professionals with AI agents. https://hr.economictimes.indiatimes.com/news/hrtech/ibm-claims-to-have-hired-more-programmers-salespeople-by-replacing-hr-professionals-with-ai-agents/121053363 ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10

  6. Mike Cannon-Brookes / Atlassian (2026-03-11), An important update on our team. https://www.atlassian.com/blog/company-news/atlassian-team-update-march-2026 ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10

  7. Duolingo via Indian Express (2025-04-30), Duolingo to stop working with contractors as part of 'AI-first' approach. https://indianexpress.com/article/technology/artificial-intelligence/duolingo-stop-working-contractors-ai-first-approach-9974797/ ↩ ↩2 ↩3 ↩4 ↩5

  8. Brody Ford / Bloomberg (2024-01-08), Duolingo Job Cuts: 10% of Contractors Laid Off With AI Features Added. https://www.bloomberg.com/news/articles/2024-01-08/duolingo-cuts-10-of-contractors-in-move-to-greater-use-of-ai ↩ ↩2

  9. Duolingo / Luis von Ahn (2025-04-28), AI-first all-hands email (LinkedIn corporate post). https://www.linkedin.com/posts/duolingo_below-is-a-note-from-our-ceo-luis-von-activity-7322560534824865792-DLdu/ ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10

  10. Aleksandra Sagan / The Logic (2026-05-04), Shopify makes cuts to its operations team in latest round of layoffs. https://thelogic.co/news/exclusive/shopify-layoffs-revenue-operations-restructure/ ↩ ↩2 ↩3

  11. Tobi Lütke / Shopify (via @tobi X account, 2025-04-07), Shopify internal AI-headcount memo. [referenced in MediaNama [47]] ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8

  12. Wade Tyler Millward / CRN (2025-02), Workday To Lay Off Employees Ahead Of Q4 Earnings. https://www.crn.com/news/ai/2025/workday-to-layoff-employees-ahead-of-q4-earnings ↩

  13. Bloomberg Law (2026-02-04), Workday to Cut About 400 Employees Focused on Customer Support. https://news.bloomberglaw.com/daily-labor-report/workday-to-cut-about-400-employees-focused-on-customer-support ↩

  14. ET / Freshworks (2026-05-06), Freshworks cuts 500 jobs; Q1 revenue hits $228 million. https://economictimes.indiatimes.com/tech/technology/freshworks-lays-off-500-employees-q1-revenue-grows-16-to-228-million/articleshow/130841734.cms ↩

  15. Sarah Bregel / Fast Company (2024-07-10), Intuit Layoffs Include 1,050 as 'Underperforming' Employees. https://www.fastcompany.com/91153816/intuit-to-lay-off-1800-employees-labels-1050-as-underperformers ↩

  16. Gyana Swain / Computerworld (2026-03-12), Atlassian cuts 1,600 jobs to fund AI and enterprise expansion. https://www.computerworld.com/article/4144218/atlassian-cuts-1600-jobs-to-fund-ai-and-enterprise-expansion.html ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  17. Marc Benioff / Salesforce (2025-04-25), FY25 Annual Report (CEO Letter). https://www.sec.gov/Archives/edgar/data/1108524/000110852425000019/salesforce_fy25annualreport.pdf ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10

  18. Salesforce, Inc. (2025-04-24), 2025 Proxy Statement. https://s205.q4cdn.com/626266368/files/doc_financials/2025/ar/Salesforce-FY25-Proxy-Statement.pdf ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13

  19. Gartner, Inc. (2026-02-03), Gartner Predicts Half of Companies That Cut Customer Service Staff Due to AI Will Rehire by 2027. https://www.gartner.com/en/newsroom/press-releases/2026-02-03-gartner-predicts-half-of-companies-that-cut-customer-service-staff-due-to-ai-will-rehire-by-2027 ↩ ↩2 ↩3 ↩4

  20. Colleen Madden Blumenfeld / Challenger, Gray & Christmas (2026-04-02), Challenger Report: March Cuts Rise 25% From February, AI Leads Reasons. https://www.challengergray.com/blog/challenger-report-march-cuts-rise-25-from-february-ai-leads-reasons ↩ ↩2 ↩3 ↩4

  21. Kate Rooney / CNBC (2026-05-04), Bret Taylor's Sierra raises nearly $1 billion months after last capital push. https://www.cnbc.com/2026/05/04/bret-taylor-sierra-fundraise-openai.html ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  22. Connie Loizos / TechCrunch (2026-05-04), Sierra raises $950M as the race to own enterprise AI gets serious. https://techcrunch.com/2026/05/04/sierra-raises-950m-as-the-race-to-own-enterprise-ai-gets-serious/ ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11

  23. Shirin Ghaffary / Bloomberg (2026-01-28), AI Customer Support Startup Decagon Valued at $4.5 Billion. https://www.bloomberg.com/news/articles/2026-01-28/ai-customer-support-startup-decagon-valued-at-4-5-billion ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11

  24. TokenRing AI / BPAS (2026-01-15), Travelers Insurance Scales Claude AI Across Global Workforce in Massive Strategic Bet. https://markets.financialcontent.com/bpas/article/tokenring-2026-1-15-travelers-insurance-scales-claude-ai-across-global-workforce-in-massive-strategic-bet ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13

  25. Simon Sharwood / The Register (2026-05-08), Cloudflare to fire 1,100 staff whose jobs just aren't AI enough. https://www.theregister.com/off-prem/2026/05/08/cloudflare-to-fire-1100-staff-whose-jobs-just-arent-ai-enough/5235536 ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8

  26. Lindsay Clark / The Register (2025-09-02), Salesforce sacrifices 4,000 support jobs on the altar of AI. https://theregister.com/2025/09/02/salesforce_4000_jobs_ai ↩ ↩2

  27. Ava Levinson / Inc.com (2026-02-10), Salesforce Reorganizes Leadership While Cutting Around 1,000 Jobs. https://www.inc.com/ava-levinson/salesforce-reorganizes-leadership-while-cutting-around-1000-jobs/91300631 ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  28. Sherin Shibu / Entrepreneur (2025-05-08), IBM CEO: AI Replaced Hundreds of Human Resources Staff. https://www.entrepreneur.com/business-news/ibm-ceo-ai-replaced-hundreds-of-human-resources-staff/491341 ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  29. Brody Ford / Bloomberg (2026-03-11), Atlassian (TEAM) CEO Announces Layoffs of 1,600, Citing AI Shift. https://www.bloomberg.com/news/articles/2026-03-11/atlassian-team-ceo-announces-layoffs-of-1-600-citing-ai-shift ↩

  30. Jaspreet Singh / Reuters (2026-03-11), Atlassian to cut roughly 10% jobs in pivot to AI. https://www.wireless.reuters.com/technology/atlassian-lay-off-about-1600-people-pivot-ai-2026-03-11/ ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8

  31. 4 Corner Resources via Challenger (2026-05-07), April Job Cuts Rose 38%. One in Four Was Blamed on AI. https://www.4cornerresources.com/job-market-news/challenger-april-2026-job-cuts/ ↩ ↩2 ↩3

  32. Colleen Madden Blumenfeld / Challenger, Gray & Christmas (2026-03-05), Challenger Report: February Cuts Plunge, YTD Hiring Falls 56%. https://www.challengergray.com/blog/challenger-report-january-job-cuts-surge-lowest-january-hiring-on-record-2/ ↩

  33. Anthropic, ServiceNow chooses Claude to power customer apps and increase internal productivity. https://www.anthropic.com/news/servicenow-anthropic-claude ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  34. Anthropic, Customer story: Block. https://www.anthropic.com/customers/block ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  35. Anthropic, Customer story: Rakuten. https://anthropic.com/customers/rakuten ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  36. HUB International + Anthropic / PRNewswire (2026-02-25), HUB International Brings Anthropic's Claude to 20,000+ Employees, Reports 85% Productivity Gains and 90% User Satisfaction. https://www.prnewswire.com/news-releases/hub-international-brings-anthropics-claude-to-20-000-employees-reports-85-productivity-gains-and-90-user-satisfaction-302696485.html ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  37. Katherine Tangalakis-Lippert / Business Insider (2026-05-07), Read the memo: Cloudflare is laying off 1,100 employees to prepare for 'the agentic AI era'. https://www.businessinsider.com/cloudflare-announces-1100-layoffs-amid-ai-focus-shift-2026-5 ↩ ↩2

  38. Salesforce, Inc. (2025-09-04), Q2 FY26 10-Q Condensed Consolidated Financial Statements. https://d18rn0p25nwr6d.cloudfront.net/CIK-0001108524/e0ddcab1-0019-4bae-b695-635df2e6ff0f.pdf ↩ ↩2

  39. Mark Haranas / CRN (2025-02), Salesforce To Lay Off Over 1,000 Employees Amid AI Hiring Spree. https://www.crn.com/news/ai/2025/salesforce-to-lay-off-over-1-000-employees-amid-ai-hiring-spree ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  40. Salesforce, Inc. (2025-04), Salesforce 2025 Proxy Supplement. https://www.sec.gov/Archives/edgar/data/1108524/000110852425000025/salesforce2025proxysupplem.htm ↩

  41. Hugh Son / CNBC (2026-02-06), Goldman Sachs taps Anthropic's Claude to automate accounting. https://www.cnbc.com/2026/02/06/anthropic-goldman-sachs-ai-model-accounting.html ↩

  42. Resultsense (2026-02-27), JPMorgan Begins Staff Redeployment as AI Reshapes Operations. https://www.resultsense.com/news/2026-02-27-jpmorgan-begins-staff-redeployment-as-ai-reshapes-operations ↩

  43. Jeremy Kahn / Fortune (2026-02-10), AI agents from Anthropic and OpenAI aren't killing SaaS—but the incumbents can't sleep easy. https://fortune.com/2026/02/10/ai-agents-anthropic-openai-arent-killing-saas-salesforce-servicenow-microsoft-workday-cant-sleep-easy/ ↩

  44. Jay Peters / The Verge (2025-04-28), Duolingo will replace contract workers with AI. https://www.theverge.com/news/657594/duolingo-ai-first-replace-contract-workers ↩ ↩2 ↩3

  45. Lauren Forristal / TechCrunch (2024-01-09), Duolingo cuts 10% of its contractor workforce as the company embraces AI. https://techcrunch.com/2024/01/09/duolingo-cut-10-of-its-contractor-workforce-as-the-company-embraces-ai/ ↩ ↩2

  46. Jennifer Korn / CNN (2024-01-09), Duolingo lays off staff as language learning app shifts toward AI. https://www.cnn.com/2024/01/09/tech/duolingo-layoffs-due-to-ai ↩

  47. Dhruv Raghav / MediaNama (2025-04-29), Duolingo To Gradually Replace Its Contractors With AI. https://www.medianama.com/2025/04/223-duolingo-contractors-ai/ ↩ ↩2 ↩3 ↩4

  48. CorePiper (2026-05-03), Decagon vs Sierra: The Enterprise AI Agent Showdown (2026). https://corepiper.com/blog/decagon-vs-sierra/ ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

Share on XShare on LinkedInAgent-readable
perea.ai Research

One deep piece a month. Three weekly signals.

Get every B2A field report, protocol update, and benchmark from real audits — published before the rest of the market sees it. No filler. Unsubscribe in one click.