# How We Hit 83% Autonomous Support (With Better Documentation)

Source: https://www.youtube.com/watch?v=ZmBwqe8wIBU
Recap page: https://rapidrecap.app/video/ZmBwqe8wIBU
Generated: 2026-02-13T05:35:36.874+00:00

---
## Quick Overview

The presentation concludes that achieving 83% autonomous support for self-serve tickets is possible by implementing a Human-In-The-Loop (HITL) process that relies on composable AI agents, where documentation is consistently updated by an AI agent (Documentation Drafter) that drafts changes based on research from another agent (Support Researcher), with all changes requiring human review before merging.

**Key Points:**
- The primary outcome achieved was 83% of self-serve support tickets being solved end-to-end by AI autopilot as of January 2026, according to Relevance AI.
- The success is attributed to a Human-In-The-Loop (HITL) process ensuring that no change goes live without human approval, involving steps: Trigger -> Research -> Draft -> Review -> Merge.
- The documentation workflow relies on composable agents, specifically using Claude Sonnet 4.5 for complex reasoning and writing, and GPT-4.1 for structured API calls (like those to Linear).
- The 'Doctor' agent orchestrates the workflow, coordinating other specialized agents like 'Support Researcher' (using 14 tools including Google Search, Firecraw, Pylon, Notion, Linear) and 'Documentation Drafter'.
- The Documentation Drafter agent automates writing by creating a branch, reading existing content, committing changes to MDX files, updating docs.json navigation, and naming the branch after the Linear issue.
- A key pattern identified is that the bottleneck for documentation work is context-switching and research, not the writing itself, which AI agents excel at assisting with.
- Future planned additions include automated release detection to trigger documentation updates based on code changes in new releases.

![Screenshot at 00:03: The title slide clearly sets the context for the presentation: 'SCALING DOCUMENTATION WITH AI AGENTS', indicating the core theme of using AI for documentation scalability.](https://ss.rapidrecap.app/screens/ZmBwqe8wIBU/00-00-03.jpg)

**Context:** This presentation, titled 'Scaling Documentation with AI Agents' by Alex Waite from Relevance AI, details the methodology and results behind achieving high levels of autonomous support resolution through advanced AI agent workflows. The talk builds upon a previous session comparing 'Auto-Pilot vs Co-Pilot' agents, focusing specifically on the mechanics that make these agents effective, particularly in the context of technical documentation updates.

## Detailed Analysis

The presentation outlines a strategy for scaling documentation maintenance using AI agents, achieving an 83% autonomous resolution rate for self-serve support tickets by January 2026. The core philosophy is 'Human-In-The-Loop (HITL) by design,' meaning that while AI handles the heavy lifting, human review is mandatory before production merges. The workflow involves several stages: Trigger (Manual, Slack, or Pylon from support tickets), Research, Draft, Review, and Merge. The orchestration is managed by 'The Doctor' agent, which coordinates specialized sub-agents. For example, the 'Support Researcher' agent uses 14 tools (including Google Search, Pylon, Notion, and Linear) to find relevant context and documentation. The 'Documentation Drafter' agent then uses this research to create new branches, commit MDX changes, update navigation files, and name the branch according to Linear issue tracking. A mixed-model strategy is employed, using Claude Sonnet 4.5 for complex reasoning and writing tasks, and GPT-4.1 for structured API calls (like interacting with Linear). The presentation also highlighted that the true bottleneck in documentation is often context-switching and research, which the AI workforce significantly mitigates. Future plans include automating release detection to automatically trigger documentation updates based on code changes in new releases.

### Introduction

- Scaling Documentation with AI Agents
- Review of prior session: Auto-Pilot vs Co-Pilot
- Focus on what makes agents work

### Why Docs Matter

- Old reason (keep docs up to date for customers) vs. New reason (Agents rely on docs as source of truth)
- Stale docs break AI

### Proof Point

- 83% of self-serve tickets solved end-to-end by AI autopilot (Jan 2026, Relevance AI)
- This success is due to the co-pilot approach and continuous documentation upkeep

### The Pattern

- Build agents that are composable, not single-purpose
- Agents should be reusable across multiple workforces (e.g., Support Researcher used by Engineering and Documentation teams)

### The Workforce

- 5 agents working together using 3 triggers and 25 tools across 65 runs in two months
- Agents include Support Researcher, Notion Assistant, Linear Assistant, Documentation Drafter, and The Doctor (Orchestrator)

### Mixed Model Strategy

- Claude Sonnet 4.5 for complex reasoning/writing, GPT-4.1 for structured API calls
- Matching the right model to the right task

### The Complete Flow

- Trigger -> Research -> Draft -> Review -> Merge
- Human approval is required before production (Always HITL)

![Screenshot at 00:00: Title slide for the presentation: "SCALING DOCUMENTATION WITH AI AGENTS" by Alex Waite, Relevance Live.](https://ss.rapidrecap.app/screens/ZmBwqe8wIBU/00-00-00.jpg)
![Screenshot at 00:03: Slide contrasting 'The Old Reason' \(Keep docs up to date for customers\) with 'The New Reason' \(Agents rely on docs as source of truth\) under the header 'WHY DOCS MATTER'.](https://ss.rapidrecap.app/screens/ZmBwqe8wIBU/00-00-03.jpg)
![Screenshot at 00:22: Slide showing the proof point: '83% Self-serve tickets solved end-to-end by AI autopilot' from January 2026 data.](https://ss.rapidrecap.app/screens/ZmBwqe8wIBU/00-00-22.jpg)
![Screenshot at 01:13: Slide detailing the 'Documentation Drafter' workflow: Creates new branch, Reads existing content, Commits MDX changes, Updates docs.json, Names branch after Linear issue, Branch -\> not merge.](https://ss.rapidrecap.app/screens/ZmBwqe8wIBU/00-01-13.jpg)
![Screenshot at 06:41: Visual representation of 'The Workforce' flow: 5 agents working together using 3 triggers, 25 tools, and resulting in 65 runs.](https://ss.rapidrecap.app/screens/ZmBwqe8wIBU/00-06-41.jpg)