Many non-technical PMs are stuck managing backlogs in tools like Jira, dependent on engineers. AI coding assistants like Claude Code empower them to contribute directly to the codebase, transforming their role from manager to builder.
Andre Albuquerque proposes a four-level progression for non-technical PMs to code with AI: 1) Start with Lovable, 2) Combine Lovable and Claude Code, 3) Transition to Claude Code and Vercel, and 4) Master multi-agent automation.
For non-technical PMs transitioning to AI coding, Lovable can serve as "training wheels" infrastructure. Write complex code in Claude, but use Lovable's simpler interface to visually QA, test, and preview changes before deploying.
When an AI-coded feature is flawed, the instinct is to patch the specific output. A more effective, long-term approach is to analyze *why* your agent system produced a bad result and improve the underlying agent, skill, or process that failed.
A powerful way to structure your AI agent system is to create a "PM agent" that acts purely as an orchestrator. It receives a task, then delegates to specialized agents (e.g., Designer, Engineer, Researcher), mimicking a real product manager's role.
AI tools collapse traditional roles. Andre suggests modern teams will consist of four archetypes: a commercial person (sales/marketing), a product builder (vibe-coding solutions), a technical scaler (ensuring reliability), and an infra/security person (protecting the system).
To avoid shipping "slop" from AI coding assistants, the solution is building robust infrastructure. Automated checks and security guardrails prevent bad code from reaching production, acting as a programmatic senior engineer for the non-technical builder.
The highest level of AI coding proficiency involves creating a "machine that builds the machine." This means developing a custom system of agents (e.g., PM, Engineer), skills, and a central `Claude.md` config that automates your unique workflow and values.
Unlike in the US, many European companies have a "Product Owner" culture where PMs act as delivery managers, lacking technical skills and decision-making power. This bureaucratic role is a major obstacle to adopting builder-centric AI tools.
For an immediate, practical start as a builder, a non-technical PM should ask engineering for access to a low-risk repository. Then, they should pick a long-neglected backlog feature and use an AI tool like Claude Code to build a first version.
Collaboration is a bottleneck during the execution phase due to dependencies. AI tools empower individuals ("teams of one") to handle execution independently, freeing the team to collaborate more effectively at the start (discovery) and end (delivery, GTM).