ARCHAngular is the right default for any regulated platform in 2026.

React's ecosystem moves faster but is less audit-friendly. NgRx + RxJS give you serializable state by default — that matters when a regulator subpoenas your deposit flow. The "Angular is slow" argument is a 2018 take that survived past its expiration date.

MFEMicro-frontends are a contract problem, not a routing problem.

Most teams adopt them for Conway-shaped reasons (we have 4 teams, we want 4 apps) and pay the wrong tax. The hard part is event contracts and shared auth/state — not Module Federation config. If your MFEs can't be deployed independently without a runtime version matrix, you built a distributed monolith.

TESTThe testing pyramid is wrong for B2C frontends.

Inverted pyramid — heavy E2E, lean units, near-zero shallow snapshots — wins when UX is the product. Unit tests catch refactors; E2E catches what the customer actually sees. On the cashier, a single Playwright trace replaces 40 brittle component tests and tells me whether revenue is at risk.

AIAI code review is already better than 80% of human reviewers for style and local correctness.

The remaining 20% — cross-system reasoning, "why is this here", judging trade-offs against constraints reviewers were never told about — is exactly what senior humans are for. Hiring should optimize for that, not for catching missing semicolons your CI also caught.

SIGNALBundle size is a leading indicator of team health, not a tech metric.

A team that ignores the 1.4MB main chunk also ignores the dead route, the four versions of lodash, and the on-call rotation. The number itself doesn't cost much; what it predicts about discipline does. -33% JS at Canon Medical was the cheapest culture audit I've ever run.

DXRxJS isn't harder than promises. It's harder than the wrong abstraction you reach for instead.

Most "just use async/await" code is a hand-rolled, buggy reimplementation of switchMap + retry + takeUntil. Pay the learning curve once; stop reinventing cancellation, backpressure, and race conditions per-feature for the rest of your career.

DSFramework-agnostic design systems lose unless the contract is shipping, not theory.

Web Components only pay off when you have >1 framework consuming them in production. Otherwise you bought distribution overhead to solve a problem you don't have. Stencil at Betsson works because Angular, React, and Ember are all actually downstream — not because someone read a blog post.

PROCIf your PRs require a meeting to merge, your architecture is unwritten.

Architecture lives in the code review template, the ADR folder, and the linter config — not in a Confluence page nobody opens. The 35-page architecture system at Betsson cut onboarding -40% because it replaced "ask Erik" with "ask the doc". The doc doesn't go on PTO.

AISpec-first is the only disciplined way to ship with LLMs and agents.

Vibe-coding generates output nobody can review. Spec-driven development inverts it: write the contract, expected behavior, and failure modes first — then the agent implements against a legible target. The spec is the review artifact, not the generated code. That is how I run the 12-agent Copilot system at Betsson: every agent gates on a written spec before touching a file. Without the spec, you have a black box you cannot audit, cannot roll back, and cannot hand off.