# 44 - From COBOL to ChatGPT: Modernizing a 110-Year-Old Tech Stack with Jeremy Uniake

Page: https://stenobird.com/podcast/cloud-currents-6702988/44-from-cobol-to-chatgpt-modernizing-a-110-year-old-tech-stack-with-jeremy-uniake
Text version: https://stenobird.com/podcast/cloud-currents-6702988/44-from-cobol-to-chatgpt-modernizing-a-110-year-old-tech-stack-with-jeremy-uniake.md
Podcast: [Cloud Currents](https://stenobird.com/podcast/cloud-currents-6702988)
Published: 2025-12-09T21:11:00+00:00
Episode link: https://traffic.megaphone.fm/LDB3017447864.mp3
Audio file: https://traffic.megaphone.fm/LDB3017447864.mp3
Processing state: not_requested
JSON: https://stenobird.com/v1/public/podcasts/cloud-currents-6702988/episodes/44-from-cobol-to-chatgpt-modernizing-a-110-year-old-tech-stack-with-jeremy-uniake
Duration seconds: 4004

## Resource

In today's episode, Jeremy Uniake, VP of IT at Thrive, shares how he's modernizing 40 million lines of COBOL while implementing AI that analyzes 10+ million calls with 98% accuracy—all at a fraction of typical enterprise costs. From managing tech across 10+ acquisitions to challenging what "AI" really means (spoiler: it's pattern matching, not intelligence), Jeremy delivers no-BS insights on legacy modernization, practical AI implementation, and why simplification is harder than anyone thinks. Learn his three-tier AI strategy, how to avoid big bang deployments, and why the future of search might be in jeopardy

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/cloud-currents-6702988/episodes/44-from-cobol-to-chatgpt-modernizing-a-110-year-old-tech-stack-with-jeremy-uniake/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/cloud-currents-6702988/44-from-cobol-to-chatgpt-modernizing-a-110-year-old-tech-stack-with-jeremy-uniake.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.