Agentic AI is repeating enterprise IT's costliest mistakes — one system cost 20x conventional architecture
DavidLinthicum · x · 2026-11-20
The author recounts a real case: a company spent roughly 20x the cost of a traditional architecture building an order intake and exception management system with autonomous AI agents, seduced by vendor pitches of '10,000 expert employees on call.' The system worked, but ROI was virtually identical to conventional software while integration, guardrails, non-deterministic testing, and operational complexity blew up the budget.
He hears variations of this weekly and maps it to IT history: SOA devolving into an architecture tax of governance paralysis, and microservices shattering simple apps into hundreds of services. Core argument: agentic AI adoption is being narrative-driven instead of ROI-driven, repeating enterprise IT's most expensive mistakes.
More from Companies & People
- Apple job listing reveals an AI agent that browses, runs terminal commands and debugs code — mark_k · 2026-11-20
- OVHcloud founder: 18 months of talks with Mistral AI collapsed, so we built our own LLM — IgorCarron · 2026-11-20
- Anthropic's 5,300 employees now out-revenue Tencent's 120,000 per head by 12x — 创业邦 · 2026-11-20
- Meta weighs dropping 2030 net-zero deadline amid AI data center buildout — Polymarket · 2026-11-20
- Signal blocks OS-level AI agents by default, calling them an existential privacy threat — mer__edith · 2026-11-20
- Solo founder ships 242 AI-assisted commits in one week while running a 7,000-member community — gefei55 · 2026-11-20