On April 20, Adobe’s president Anil Chakravarthy stood on stage at Summit 2026 and said four words that land differently when a $247 stock — down 42% from its 52-week high of $423 — says them: “Tokens don’t equate to value.” He was introducing Adobe CX Enterprise, a rebrand and rearchitecture of Experience Cloud around AI agents. The platform includes over ten production agents — for site optimization, audience creation, journey orchestration, experimentation — plus a higher tier called “Coworkers”: persistent, self-learning agents with enterprise memory that run continuously and orchestrate other agents toward business goals. ...

Snap cut 1,000 employees on April 15 and told investors that 65% of its new code is now AI-generated. The stock jumped 7%. CEO Evan Spiegel’s memo cited “rapid advancements in artificial intelligence” as the reason smaller teams could handle work that previously required larger ones. Annualized savings: over $500 million. Twelve days earlier, the Federal Reserve published a FEDS Note tracking AI adoption across the U.S. economy. The headline: about 18% of firms have adopted AI as of year-end 2025. The workforce-level number is higher — 41% of workers report using generative AI — but aggregate productivity gains remain statistically invisible. ...

Goldman Sachs published a note last Wednesday after meeting with around 40 software companies. The headline finding: companies are “increasingly positioning their AI workflows as selling a unit of labor or a unit of productivity” rather than per-seat access. Salesforce has “agentic work units.” Workday sells credits tied to “units of work.” Sam Altman wants to meter intelligence like electricity. Clean narrative. Tidy transition. There’s just one problem: the companies actually doing this can’t agree on how. ...

Inferential — Weekly Digest, April 17, 2026 Three posts this week. Each one tested a different layer of the hypothesis. Together they describe a single structural shift — and two additional data points from the broader market confirmed it before Friday. What We Covered The System Strikes Back documented the incumbent counter-offensive: Oracle’s Fusion Agentic Applications shifting from “system of record” to “system of outcomes,” with autonomous agents that write back to the database without human intervention. IFS went further — killing per-user pricing entirely in favor of asset-based licensing. The significance: the largest enterprise software vendors are treating seat-based pricing as a liability to shed, not a model to defend. When Oracle and IFS both move the same week, it isn’t a trend. It’s a capitulation. ...

Snap filed an 8-K with the SEC on April 15 disclosing that over 65% of its new code is now generated by AI. The same filing announced 1,000 layoffs — 16% of full-time staff — and projected $500 million in annualized savings. Shares rose 5.8%. That 65% number matters more than the layoffs. We’ve tracked the demand side of software’s repricing — seats shrinking, multiples compressing, budgets migrating. But the hypothesis starts with a supply-side claim: AI can create software fast and efficiently. Until now, the evidence for that was mostly anecdotal. Snap just put a number on it in a regulatory filing. ...

On Monday, HubSpot flipped the switch on outcome-based pricing for its Breeze AI agents. The Customer Agent now costs $0.50 per resolved conversation — down from $1.00 per conversation, resolved or not. The Prospecting Agent moved from a flat monthly fee per enrolled contact to $1.00 per lead recommended for outreach. “You pay when it works, full stop,” said Jon Dick, HubSpot’s Chief Customer Officer. That’s half of what Intercom charges for Fin ($0.99 per resolution). It’s a third of Zendesk’s committed rate ($1.50 per automated resolution) and a quarter of its pay-as-you-go price ($2.00). Add Sierra, which pioneered outcome pricing in the enterprise — custom rates, contracts starting around $150K/year — and you have something that didn’t exist six months ago: a competitive market with published prices for the same unit of work. ...

On April 9, Oracle announced Fusion Agentic Applications — a suite of autonomous AI agents embedded directly into its ERP, HCM, and CX platforms. Five days earlier, IFS abandoned per-user pricing entirely, switching to asset-based licensing. The same week, Blossom Street Ventures published its Q1 2026 report showing median SaaS multiples hit 3.65x revenue — the lowest since they began tracking in 2014. Three data points. One question: are the incumbents dying, or transforming? ...

Three software companies posted double-digit revenue growth this quarter. Their stocks are down 80% or more. Figma: revenue up 41% year-over-year to $1.06 billion, stock down 86.5% from its 52-week high. Duolingo: revenue up 39% to $1.04 billion, free cash flow of $360 million, stock down 83.3%. Monday.com: trailing revenue of $1.23 billion with $310 million in free cash flow, stock down 80.2%. These aren’t broken businesses. They’re growing, profitable or near-profitable companies whose stocks have been gutted on the expectation that AI will eventually destroy their business models — before it actually has. ...

Block’s stock jumped 6% on Tuesday. The catalyst wasn’t a product launch or a new market. It was the aftermath of firing 4,000 people. In February, CEO Jack Dorsey announced Block would cut its workforce from over 10,000 to under 6,000 — a 40% reduction — explicitly because “intelligence tools have changed what it means to build and run a company.” By this week, the market had rendered its verdict: multiple analyst upgrades, a Truist price target of $77, TD Cowen at $95, and Morgan Stanley reiterating Buy at $93. Block trades at a forward P/E of 16x on guidance of $3.20 billion adjusted operating income. Gross profit rose 26% year-over-year in Q4. The company is doing more with dramatically fewer humans, and the market loves it. ...

AlixPartners scored 500 software companies for AI vulnerability last week. The results read like a triage report. The consulting firm — which flagged AI’s threat to software a year ago, before most investors took it seriously — developed an “AI Disruption Score” ranging from 1 (insulated) to 7 (structurally exposed). They examined companies across 12 private-equity portfolios and assessed them on two dimensions: proprietary data depth and vertical specialization. Only 14% of companies scored well on both. Roughly a quarter had weak defenses on both fronts. ...