Forbes: "Enterprise Brain" Replaces AI Agents as the Next Architecture
Sandy Carter's deep dive on what comes after AI agents: the enterprise brain. Microsoft launched its IQ family (Work IQ, Fabric IQ, Foundry IQ, Web IQ) as a shared context layer across all agents. UnifyApps co-founder Ragy Thomas frames it biologically — enterprises spent a century building the body (supply chains) and nervous system (cloud software), but still can't think. The constraint isn't model quality; it's architecture. Only 28% of AI use cases fully succeed, per Gartner. The fix isn't a better model — it's a brain that lets all agents learn collectively.
Read on Forbes →Salesforce Agentic Enterprise Index: Agent Deployments Nearly Tripled YoY
The numbers are striking: average agents activated per org up nearly 3x over fiscal year. Average time from creation to use: 2 days, down 53%. Agents now execute 6 skills on average (up from 2). During peak retail, agents scaled to 9 skills — a 350% increase to handle complex holiday demand. Retail AI agents drove 4x higher online sales growth. Financial services deploys the most sophisticated agents at massive scale, proving complexity and volume can coexist.
Read on Salesforce →Forbes AI 50: Anthropic Hits $30B Run Rate, OpenAI Crosses $25B
The revenue proof is in: Anthropic crossed $30B annualized revenue in April 2026; OpenAI surpassed $25B by late February. Together they account for 80% of the $305.6B in total VC funding across all 50 AI 50 companies. Three companies from last year's list got acquired — including Google paying $2.4B to acquihire Windsurf's cofounders. The message: enterprise AI has moved decisively past the hype phase into real revenue at scale.
Read on MarketScale →SoftBank Unveils AGENTIC STAR at AI WAVE SHOW 2026
At Taiwan's AI WAVE SHOW, SoftBank unveiled AGENTIC STAR — its enterprise-grade AI agent platform — for the first time outside Japan. Meanwhile, L&T Technology Services announced AgenticIQ, an end-to-end agentic AI platform for engineering and manufacturing. Google's Gemini Spark agent can now operate desktop Chrome using saved passwords to handle tasks like booking property viewings. The race to build production-ready agent platforms is accelerating globally.
Read on PR Newswire →Gartner: 40% of Enterprise Apps Will Feature AI Agents by End of 2026
The shift is happening faster than expected: Gartner now projects 40% of enterprise applications will feature task-specific AI agents by end of 2026 — up from less than 5% a year earlier. Separately, Gartner predicts 33% of enterprise software will include agentic AI by 2028. The implication: AI agents are becoming table stakes for enterprise software, not a differentiator. Companies that can automate most repetitive work with no-code agents are winning.
Read on TechRev →💡 PMM Takeaway
Read the Forbes piece on "Enterprise Brain." The insight that only 28% of AI use cases succeed — and that the constraint is architecture, not models — is the story. Microsoft's IQ family and UnifyApps are racing to build the shared memory layer that lets all agents learn collectively. For PMMs, this means: stop messaging around "AI agents" and start messaging around context, memory, and enterprise-wide intelligence. The companies that solve the coordination problem win. Meanwhile, Salesforce's data shows agent adoption is real and accelerating — 3x deployments, skills tripling, retail seeing 4x sales lift. The early adopters are past pilot phase.