Project Documentation

Creating Documentation for a fictional Gmail feature “Voice Command”

Introduction

The project will demonstrate how AI can be effectively used in technical writing.

It will also demonstrate that documents still require human oversight, creativity, and technical understanding.

Starting Key Points

When creating documentation for a fictional product:

The whole concept is on you, but you are free to use your imagination as long as it makes sense (no Jira tickets, developers, product managers, etc.)
The easier part compared to a real product – you are not dependent on others, there is no reading, watching material, no consulting stakeholders
The more difficult part – you cover the product manager and business analyst’s roles, it’s time-consuming, and serious analyzing is involved

User Guide

Starting point – I started the project by asking AI to create a well-structured user guide, which it did. The document structure along with some content that I got in a second was an excellent starting point, saving me a lot of time. The content also gave me some ideas.

Work on the document – Then, I worked on the content while also improving the overall idea of the feature:

It had to be logical, functional and useful to future users
It also had to be intuitive, flexible and easy to use
With implemented AI, good analytics and synonyms list, voice commands would gradually evolve

The document and the concept finished – Having the whole concept of the Gmail voice command completed with this user guide, I was ready to work on other documents.

It turned out I needed to update it several times because I got new ideas.

PDF Document

Full Screen Download

Feature Overview

Using the user guide as a base, it was easy to create a feature overview document. The AI-generated content was satisfactory, requiring almost no adjustments. I changed the language slightly.

PDF Document

Full Screen Download

Technical Documentation

As someone with no formal technical education and unable to rely on developer colleagues, I needed AI’s help the most for the technical documentation. Based on the previous two documents, AI generated detailed technical content, including flowcharts (using Mermaid syntax) and code snippets. These last two elements helped me maintain some control, because it was easier to check the accuracy and logic with the flowcharts and codes than with other technical details.

In short, I reviewed the content to the best of my ability, investigated technical aspects, and shortened the document where necessary.

In a standard procedure, the next step would be to send it to someone for review, but not this time.

PDF Document

Full Screen Download

Frequently Asked Questions (FAQ)

I used AI to generate 10 FAQs for the feature. Working on this document gave me new ideas for the entire feature, leading me to update already finished documents. As I mentioned earlier, this process of updating documents happened several times.

PDF Document

Full Screen Download

Release Notes

I wrote the release notes for external users, incorporating improvements announced in the previous documents.

As someone who has written tens of thousands of release notes and created templates for them, this document was not easy because I had no idea who the end-users would be, what data to include, and what form would be best (tabular or textual). So, I assumed they were for external use and created them using simple and clear language that anyone could understand.

PDF Document

Release Notes Download

Conclusion

Throughout the process, I did my best to ensure consistency.
The language in all documents was simple and clear, except when using technical terms was necessary.
The easy part of this project is that it was for a fictional product.

Creating content for real products by gathering information from various sources, which is a normal part of a technical writer’s job, can be frustrating and exhausting. I strongly believe that embracing AI assistance can significantly speed up our work and take it to the next level.