Watch the idea
The concept
Builder
RAG, agents, evals, APIs
Deployer
Sit with the customer and install
Product
AI PM · what not to automate
Domain + AI
A skill inside marketing, HR, ops
Four doors in 2026
Builder: AI Engineer, LLM Engineer, MLOps — ship retrieval, agents, evals, APIs. Deployer: consultant and forward deployed — sit with the customer and install the workflow. Product: AI PM. Domain: marketing, ops, HR, finance with AI as a skill inside the role.
Prompt engineering is a skill, not a title
Standalone prompt-engineer titles narrowed. The craft did not. Every door still pays for people who can specify role, goal, constraints, examples, and output shape — and who know when not to automate.
The market hires proof
Certificates and tool names without a shipped artifact are weak. A RAG assistant with evals, a case study with before/after time, or a portfolio that a hiring manager can click — that is the map's destination.
Worked example
Maya and Arun
Maya is a marketer. She does not become an engineer. She ships a content assistant with a brand voice file and a human edit gate, then puts the time-saved number on her resume. Arun is a CS graduate. He ships the retrieval pipeline and the eval set. Both are hireable. They picked different doors.
Picture to keep
Four doors
- 1Build systems
- 2Install them for clients
- 3Product & strategy
- 4Domain + AI