5 Work Principles from Claude Code’s Team Every Engineer Should Adopt

1. Planning Changes

Fiona Fung explains that traditional software processes focused on extensive upfront planning because coding time was expensive. In Claude Code’s fast‑iterating environment, a six‑month roadmap became obsolete within three months, leading to a Just‑In‑Time (JIT) planning approach: plan only enough at the moment, discuss directly in PRs or prototypes, and iterate based on real feedback.

The team eliminated lengthy design documents, treating them as theatrical rituals, and shifted to prototype‑first development, only writing documentation after a usable implementation exists.

2. Automation Changes

Every recurring task triggers the question, “Can this be automated?” Fiona shares a personal habit of turning a daily manual summary of customer feedback into an automated background job, freeing her to drink coffee while the system runs. The team treats any activity repeated more than three times as a candidate for AI‑driven automation, noting that automation cost is now near zero.

Examples include a Claude‑generated hook that pulls the latest GitHub code before opening a project, which runs in minutes without manual effort.

3. Code Review Changes

Facing the question “How can we keep up with review speed?”, the team adopted a “trust but verify” model. Claude handles style checks, linting, PR feedback, bug detection, and test generation—tasks that previously consumed 60‑70% of review effort—while human reviewers focus on legal compliance, security, and nuanced product decisions.

The balance between trust and verification is dynamic; today’s human‑only checks may be fully automated tomorrow.

4. Team Role Changes

Roles have become fluid: product managers write code, engineers draft content and design, and Claude drafts initial copy for bug‑fix releases, with humans providing final judgment. This reduces hand‑off delays and eliminates stale bottlenecks.

Hiring now emphasizes two traits: a product‑sense creative builder who can quickly prototype, and a deep‑system engineer who handles the subtle errors that only humans can catch. Fiona stresses that “taste is scarce, typing is not.”

5. Driving Team Change

Fiona distills three core principles: keep the organization flat and allow fluid movement between tasks; let Claude handle what it can to free humans for harder work; proactively identify and retire obsolete processes.

She illustrates this by questioning the value of a weekly status meeting that turned out to be unnecessary, leading to its cancellation. The broader lesson is that as AI penetrates deeper, many legacy steps become automatable and should be eliminated to avoid formalistic waste.

Fiona ends with three open questions for the community: the necessity of separate iOS/Android teams, the limits of fully automated review, and how to maintain confidence as roles blur. She concludes that every major technology shift forces a complete redesign of planning, knowledge management, review processes, and talent structures.

Overall, the guiding habit is simple: automate repetitive work, discard useless processes, and delegate judgment to AI when appropriate—progressing step by step without rushing.