Skip to content

Case study · Production

Agentic DevRel Content Pipeline

An autonomous five-agent LLM pipeline that generates Harper's developer relations technical content, cutting delivery time by more than 60%.

Role
Head of Developer Relations
Timeline
2025 – 2026
Stack
  • LLM orchestration
  • AI agents
  • TypeScript
  • DevRel

Overview

Developer relations runs on technical content: tutorials, guides, example apps, and enablement material that stays in sync with a fast-moving product. At Harper, producing that content by hand was the bottleneck.

As Head of Developer Relations, I architected and deployed an autonomous five-agent LLM pipeline that streamlines how technical content is generated, so the team could spend its time on review, community, and strategy instead of first drafts.

Approach

  • Specialized agents, not one big prompt. The work is split across five cooperating agents, each responsible for a distinct stage of producing a piece of content.
  • Built around review. The pipeline produces work for humans to approve, keeping technical accuracy and Harper's voice under editorial control.
  • Grounded in the product. Output is tied to real Harper APIs and example code, the same material behind Harper's tutorials and templates.

Outcome

  • 60%+ reduction in content delivery time and turnaround.
  • More capacity for the rest of the role: developer education, community strategy, technical onboarding demos for enterprise partners, and hosting the Harper SELECT * podcast.