AI-native companies treat agents as team members — onboarding, proficiency levels, and skill codification¶
Insight: According to Peter Yang's reporting, AI-native companies like Linear, Ramp, and Factory operate with agents as first-class team members rather than tools. Linear assigns agents to projects and holds them accountable alongside humans. Ramp defines four AI proficiency levels (L0-L3) and makes AI competency a hiring requirement. Factory ($300M valuation, 55 people) encodes expert knowledge into reusable "skills" — markdown files that both agents and employees invoke.
Detail: Linear's four-stage agent workflow: (1) agents summarize customer conversations and auto-create deduplicated issues, (2) agents pull data-backed insights to refine specs, (3) agents break specs into tickets and auto-route, (4) small features go directly to agents while complex work uses Claude Code with full issue context. Head of Product Nan Yu: "The scope of what agents can handle is expanding every quarter." Ramp's L0-L3 framework: L0 = occasional ChatGPT use (self-select out), L1 = experiment with GPTs, L2 = automate actual job components, L3 = build AI infrastructure accelerating others. Factory's agent-ready codebase has five maturity levels with "Standardized" as the realistic first target.
Sources
- Peter Yang — "Your New Job Is to Onboard AI Agents" (2026-03-04)