Case Study
AI-Enabled Process Documentation
Turning historical documentation into an indexed knowledge foundation and AI-supported guidance.
The Problem
Marketing processes depended on knowledge held by a small number of people. Requesters and new team members needed guidance on intake, service levels, lifecycle rules, scoring, handoffs, and attribution, often while working in another tool.
Historical documentation contained the knowledge, but it needed to be collected, translated, and organized into an accessible reference. The opportunity was to use AI throughout that transformation, then make the resulting guidance available through support agents.
AI from Knowledge Discovery to Process Support
I used AI to review the historical documentation and convert it into accessible Markdown files with a documentation index. Organizing that legacy material created a foundation for developing and maintaining process guidance.
The indexed material supported the buildout of a comprehensive documentation suite of more than 100 pages across 14 chapters, covering intake, service levels, lifecycle, scoring, handoffs, and attribution.
I built corresponding AI support agents in Glean and Asana, connecting the documentation foundation to tools people could use for process questions. AI supported both the transformation of historical knowledge and access to the resulting guidance.
The adoption strategy starts with a defined need: help people get a useful process answer when they need it. Further development is guided by use and answer quality, with process ownership and review remaining with people. The same task-first approach guides my AI-assisted software development: choose work by value and feasibility, then review the output.
What Changed
- A maintained reference gives operating guidance a place beyond individual conversations.
- Marketers can look up process guidance through support agents in Glean and Asana.
- Human owners remain responsible for operating rules, content review, and corrections.
- Usage and answer quality guide further development; agent availability alone is not evidence of adoption.