06-reference

data engineering central dave langer cobol to copilot

2026-07-01·reference·source: Data Engineering Central·by Daniel Beach (host); Dave Langer (guest)

From COBOL to Copilot: 30 Years of Data, BI, and AI with Dave Langer

Source: Data Engineering Central Podcast Guest: Dave Langer — LinkedIn, The DIY Data Scientist (Substack), author of Python and Excel Step-by-Step Format: ~57-minute video/podcast interview

Summary

A 30-year retrospective on the data industry from someone who started on a Commodore 64 and coded COBOL on IBM mainframes, went through enterprise architecture, Microsoft's Xbox division, machine learning, startup leadership, and now builds one of the largest personal brands in data.

What We Cover (from the newsletter)

Key Framing

Dave's central thesis: the biggest problems in data haven't changed — organizations still struggle with data quality, governance, trust, and self-service. The tools change every 5 years; the underlying organizational and analytical challenges do not.

On Copilot/AI as tools: these should be viewed as partners, not replacements. Strong analytical skills matter more than ever because AI amplifies the practitioner — someone who can't think analytically won't be improved by a Copilot, they'll just produce wrong output faster.

On self-service analytics: still largely a failure in most organizations. Semantic layers and proper dimensional modeling are prerequisites for AI to be useful in analytics — garbage in, garbage out, and AI doesn't fix that.

Why this is in the vault

This is a signal-dense 30-year longitudinal view of the data industry from a practitioner who lived through every major wave — mainframe → BI → analytics → data science → AI. It provides grounding for two things RDCO needs:

  1. Client positioning: When phData or RDCO clients ask "what does AI change about data?", the answer is "less than you think at the organizational layer, more than you think at the tooling layer." Dave's framing validates the RDCO positioning that data fundamentals (modeling, governance, trust) are still the bottleneck.

  2. Narrative for workforce transformation: The COBOL → Copilot arc is exactly the kind of human story that resonates with enterprise data teams who feel anxious about AI. It's usable as a rhetorical frame in client conversations: "someone who coded COBOL in the 80s is still relevant today because they learned to learn."

Mapping against Ray Data Co

Strong mapping on three axes:

  1. phData client conversations: The "core data problems haven't changed" thesis is a powerful counter-narrative to the "AI will fix our data problems" oversell. Useful when scoping discovery with clients who think AI is a shortcut around data governance.

  2. Semantic layers + AI readiness: Dave's point that semantic layers and governance are prerequisites for AI in analytics maps directly to phData's data platform work. Clients who haven't done the foundational work are not AI-ready — this is a recurring scoping pattern.

  3. RDCO personal brand / Ray's own arc: Ray has his own version of this story (analytics → DSA → AI COO). The "tools change, analytical thinking doesn't" frame is directly applicable to Ray's positioning as a practitioner who bridges data fundamentals and AI implementation — not just an AI hype merchant.

Weaker fit: The personal branding / book-writing angle (Dave's independent creator path) is directionally interesting but not an active RDCO priority right now.

Related