Alter Everything 198: Best Practices for Integrating AI Into Your Alteryx Workflows
27 snips
Nov 19, 2025 Join Andrew Merrill, an Alteryx product specialist with a background in math and data science, as he dives into integrating AI into Alteryx workflows. He shares best practices like using feedback loops and RAG architectures, while highlighting common pitfalls such as token overuse. Learn how to leverage Alteryx Co-Pilot for efficient workflow design and explore innovative use cases like weekly status report matching. Andrew emphasizes making AI workflows agentic, accessible, and structured to maximize their potential, transforming mundane tasks into higher-value work.
AI Snips
Chapters
Transcript
Episode notes
Build An LLM Feedback Loop
- Put a Gen AI tool on the canvas and iterate by judging outputs with a second LLM or an iterative macro.
- Feed outputs back in as a feedback loop to recursively improve results.
Use LLMs For Routing And Branching
- Use an LLM to classify inputs into simple categories and route records down different workflow branches.
- Let a filter tool act on the LLM's single-category output to enable automated routing.
Add Focused Memory (RAG) To Requests
- Implement a RAG-style memory: classify inputs, append only the relevant context, then run a second LLM with that focused memory.
- Avoid sending all context to prevent muddied results and hallucinations.