The Joe Reis Show

The episode contains a useful, structured framework (four morbid symptoms, three skill tiers, signals to watch) and a few genuinely sharp observations, but is diluted by repetition, throat-clearing, and some obvious industry-watcher takes that circulate widely. The insight-per-minute ratio is moderate for a 32-minute solo monologue.

“vendors are selling some version of the future which we haven't yet determined. Um, but the practitioners are maintaining the present and the business is still waiting for results and often living in yesterday's world”

“you can sell around the data team, uh, but you can't necessarily deploy around the data team or engineering teams”

The Gramsci 'interregnum' framing is borrowed (and attributed to Tristan Handy), the bundling/unbundling cycle is well-worn, and much of the skills-evolution narrative is familiar territory. The contrarian note that consolidated platforms face a new unbundling threat from AI-generated apps and raw Iceberg adoption is the freshest angle in the episode.

“you could be looking at right now. You're looking at you know, obviously bundling but you, you're looking at the early stages of, I'd say the next unbundling”

“I was talking to a team the other day, they're ripping out tableau, uh, and replacing it with react apps”

This is a solo monologue by Joe Reis, co-author of Fundamentals of Data Engineering, with genuine practitioner and author credentials and evident access to industry principals like Tristan Handy. The absence of any guest limits the ceiling on this dimension, and Reis himself has increasingly become a conference/media figure rather than a pure operator.

“last month I was chatting with uh, Tristan Handy, uh, from uh, co founders of dbt, and he called what's going on right now, uh, the interregnum”

“As I've always said, that's why I wrote a book on fundamentals. That's why I wrote a book on mixed model arts. These are technology agnostic”

The episode earns above-average specificity marks through named acquisitions with closing dates, an ARR figure for the FiveTran-dbt combination, and cited job-market statistics from DataCamp and Indeed. Anecdotal evidence (Tableau being ripped out for React apps at multiple large companies) is concrete even if unverified.

“the original combination was um, described as approaching what, 600 million in ARR”

“Data Camp, for example, they had some analysis. Roughly 2 million job postings. Um, and it seems like data and AI roles are up 80 point, uh, 2% year over year, with AI engineering roles up, uh, 255%”

As a solo monologue there is no host-guest dynamic, no follow-up questions, and no productive disagreement; the 'craft' that exists is structural organization of talking points, which is decent but not a substitute for genuine conversational tension or probing. The episode drifts and loops, and ends with a lengthy self-promotional outro.

“Anyway, that's the rant for today. Kind of a long rant. He's 30 minutes.”

“Um, I don't know. That would require tracking.”