Linda Spiller | Portfolio
AI workflow projects
-
Project: Experiments with AI to improve technical writing workflows
-
Role: Technical Writer and Technical Editor
-
Type: Flowchart, schema, and Postman testing workflows
-
Format: Online
Overview
I actively explore AI tools and their potential to enhance documentation workflows. I’ve evaluated ReadMe’s Owlbot and LLM integrations to improve user experiences. I also use tools like Grammarly and QuillBot to support clarity and consistency. I've completed courses in AI, generative AI, Copilot, and AI writing tools, and prompt creation.
Outside of work, I research AI-driven formatting tools and extensions to optimize Markdown-based environments such as Visual Studio Code. I’ve experimented with generating schemas from code samples to improve modularity, and my next step is to test generating code from schemas. With a grounding in prompt creation and AI ethics training, I’m prototyping smarter docs while aligning with responsible tech practices.
Application is no longer available.
Application is no longer available.
Create a Postman guide and build a collection of test snippets
July 2025
Tools: Copilot and Gemini
Key tasks I did:
-
Set up a personal Postman account to explore API testing workflows.
-
I tested several public API endpoints using curated code samples with the assistance of Copilot.
-
Validated response structure and status codes using Postman’s test scripting feature.
-
Evaluated test snippets and expanded testing options using Gemini.
-
Compiled notes and code snippets into documentation assets based on real-world request results.
What I learned:
-
When to use Postman tests in an API documentation workflow.
-
Testing options for checking the validity of code samples.
-
Methods for managing documentation drift: checking if a parameter still exists and comparing two similar code samples with the code in an API.
-
Postman’s tests are mini logic checks - they are structured assertions, so a collection of test snippets is an ideal way to create a set of test templates.
Impact:
-
Learned a valuable technical writing skill and created guides for Postman test snippets and processes for future use.
Application is no longer available.
Test Draw.io's Smart Template feature for creating Mermaid.js diagrams
June 2025
Tools: Draw.io (Smart Template feature) and Mermaid.js Live Editor
Key tasks I did:
-
Researched current Draw.io features that are AI-based.
-
Created an outline of the process I wanted to diagram.
-
Wrote a prompt for a sequence diagram.
-
Used the Smart Template AI feature to generate the diagram and then used Draw.io tools to adjust the colors and formatting.
What I learned:
-
How to use several tools in Draw.io to generate Mermaid.js diagrams and similar flowcharts using an AI.
-
Draw.io can also create diagrams using existing Mermaid.js code.
-
Learned how to troubleshoot prompts.
Impact:
-
Added a new tool that I can quickly use to create diagrams and flowcharts to my technical writing toolbox.
Application is no longer available.
Evaluate Copilot's ability to successfully create code for a mermaid.js sequence diagram
September 2024
Tools: Mermaid.js Live Editor, Copilot, Wix (blog), and LinkedIn.
Key tasks I did:
-
Researched Mermaid.js.
-
Located resources and samples to include in the prompt.
-
Drafted prompts to guide Copilot.
-
Tested prompts that defined a persona to guide Copilot.
-
Tested code solutions provided by Copilot in Mermaid Live Editor.
-
Drafted a blog article and LinkedIn post documenting the results.
What I learned:
-
How to use personas and detailed prompts to guide AIsHow to test AI responses and troubleshoot unexpected answers.
Impact:
-
Learned about using AI to generate code for Mermaid.js. Learned how to write detailed AI prompts.
Application is no longer available.
Application is no longer available.
Use AI to develop a workflow for creating a schema from a code sample
Tools: Copilot and Gemini
June 2025
Purpose: Create an AI-based workflow that will help speed up creating a schema table for projects that require schemas for multiple endpoints when a product team has provided code samples but not schemas.
Key tasks :
-
Prompt 1: Create prompts for both Copilot and Gemini to create an API schema from this request code sample, using a table or dot notation, and identify which parameters are arrays, strings, integers, and other types.
-
Prompt #2: Compare the results with the documentation found on this guide page and this reference page.
-
Prompt for Gemini: Add a column with sample values.
-
Produce a summary of the discussion and options provided.
Results:
-
Copilot quickly created schemas in dot notation and a table in Markdown.
-
Gemini was concise but took longer to come up with a response. Its schema matched the ones I had already developed.
-
Gemini required a code block with backticks to indicate the language. Backticks weren't necessary for Copilot.
-
Gemini's table format was simpler to read and could be saved in Google Sheets or copied into a text editor.
-
Copilot did a deep dive into the subject and offered several solutions.
-
Copilot developed an OpenAPI schema (YAML) that is compatible with ReadMe. Additionally, it recommended importing the same OpenAPI YAML into Stoplight Studio based on our prior exchanges. It also proposed a common model (component modularity) format based on this schema.
Impact:
-
Added a new workflow that I can use to create schema tables from code samples for large projects.
-
I evaluated the strengths of several AIs. For example, I learned that Copilot is great for learning new methods and exploring topics. Gemini stays on task but requires prompts that follow specific formatting requirements.