Skip to main content

November 2026 Pipelines Qrew Meetup Recap

  • Nov 20, 2026 5:16 AM UTC
  • 0 replies
  • 12 views

EstherLaVielle
Forum|alt.badge.img+12

Thank you to everyone who joined the November 2026 Pipelines Qrew meetup on Wednesday November 18th. Here’s what we discussed:

Blake Harrison started with a practical problem and free data and Pipelines to solve it. His team needed the county for each mineral owner's address before contacting the right courthouse.

  • Instead of paying for a mapping service, he used the free U.S. Census Bureau geocoder. When an address is saved without a county, a pipeline sends it to the Census and adds the matching county to the record. If nothing matches, it creates a follow-up for the research team.
  • Blake's slide deck is attached to this community post, so you can download it and walk through the step-by-step setup. You can also try the Census geocoder at https://geocoding.geo.census.gov. Keep in mind the free use guideline of about 10,000 requests a month.

Wes McAda built a "smart shopper" for a very specific need: a sousaphone without a bell.

  • Searching eBay by hand was taking forever, so he built an app that pulls eBay listings into Quickbase every day using the free Browse API: https://developer.ebay.com/api-docs/buy/static/api-browse.html.
  • A second pipeline sends the recent listings to an OpenAI chat step that checks for a match and returns a confidence level. Wes paid for his own OpenAI credentials, so keep an eye on token costs.

Qrew tips from the meeting you can apply today:

1. Add trigger conditions so the pipeline only runs when it's needed. Blake's pipeline runs only when the county is empty, the address isn't international, and it isn't a PO box. This keeps unnecessary calls and costs down.

2. Build a fallback for no-match results. Rather than failing silently, Blake's pipeline creates a follow-up for a researcher, and the county search can be triggered again manually if needed.

3. Page through API results with a recursive pipeline. Wes's pipeline uses limit and offset values and calls itself until the browse response has no more pages, so the full result set gets processed.

4. Ask the AI step for structured JSON output. Wes defined a JSON schema for the chat completion, so each result comes back with a clear match answer and confidence level that you can map to fields. He also logs token usage so you can track costs.

Missed yesterday's meeting? Here's the recording

Thanks again to Blake and Wes for sharing such creative work. We hope to see you at the next Qrew meetup!