Silver-bullet architecture

Reuse via shared services — in theory. In practice: central buses and political ownership fights.

$ ESBs, governance boards, and "service portfolios" burned seven-figure programs. Many orgs got distributed monoliths with worse latency and a permanent integration team.

Independent deployability as religion. Correct for some; ruinous cargo-cult for most. Consolidation and monolith return are the 2020s epilogue.

$ Teams split 4-person products into 40 services, then hired SREs to babysit the mesh. Many later consolidated — after years of distributed tracing invoices and on-call trauma. See case studies on premature microservices and big-bang cutovers for named failure modes.

Functions as the unit of deploy. Brilliant for spikes and glue; awkward as a whole product architecture.

Language-agnostic distributed objects. Complexity scaled faster than value for most shops.

$ Enterprises spent years on IDL, ORBs, and vendor lock-in for distributed objects that SOAP and REST later made look quaint — after the invoices cleared.

Enterprise web services with ceremony. Interoperability theater with XML envelopes.

$ Banks and governments sunk fortunes into WS-Security, WSDL toolchains, and vendor stacks — then quietly rebuilt on REST/JSON while still paying maintenance on the old bus.

Container orchestration that became the cloud's OS. Overkill for many; unavoidable for many others.

One deployable with clear boundaries. The unfashionable default that keeps shipping while microservices argue about ownership.

EJB-heavy stacks taught a generation that XML config was architecture. Spring and lighter runtimes absorbed the survivors.

Clients ask for exactly what they need — and sometimes invent a second backend team to make that safe.

Pre-render + CDN + APIs. The brand cooled; the pattern (static where possible) stuck.

Decentralize data ownership like microservices for warehouses. Real org problem; often a logo refresh for the same broken pipelines.

$ Consultancies sold domain ownership and federated governance while teams still could not agree on a customer ID. Slideware outpaced catalog tooling by years.

Retrieval-augmented generation as the answer to every knowledge problem — then "just stuff the window" as the counter-fad. Often right; often Postgres with pgvector would suffice.

$ Teams stood up vector pipelines, chunking strategies, and rerankers before asking if fine-tuning or a SQL query would answer the question. Retrieval infra became the product.

Cheap model drafts; frontier model escalates. The boring cost-control layer that survived the agent hype.

Slap an LLM on the homepage and call it AI-native — "we have an app now," 2010 edition.

$ Homepage chat boxes billed as AI-native platforms while the roadmap stayed FAQ search with a spinner. Acquisition decks outpaced retention.

Million-token windows as a substitute for retrieval design. Stuffing PDFs until the model shrugs.

$ "Just put it in the window" replaced chunking, RAG, and information architecture until token bills and lost-in-the-middle hallucinations arrived.