Brian Pasch Podcast

There are a handful of genuinely useful ideas - AI fragmentation across dealership vendor stacks, the probabilistic vs. deterministic distinction, and the OEM re-engagement data - but they're buried under repetition, mutual compliments, and generic 'AI augments humans' filler that pads the runtime significantly.

“we already have system sprawl and we already have data sprawl...now we're talking about adding AI tools on top of tools and on top of systems. None of them are talking to each other”

“leads would come in, consumers would engage day one...About six to seven weeks we saw 40% of this audience re engage out of nowhere. And then in about two to three weeks, purchase”

The 'engagement operating system' framing and the discovery that a cohort went silent for 6-7 weeks due to liquidity constraints show real original thinking, but most of the episode recycles widely circulated AI vendor talking points - 'hire don't buy,' 'augment don't replace,' 'it learns over time' - without meaningful first-principles challenge.

“I did not break your process. I revealed your process was already broken”

“that particular audience was liquidity constraint...They started shopping in Thanksgiving”

Albert Thompson has legitimate operator credentials - Carfax, AutoTrader, Trade Rev, a first-party identity layer exit in early 2024 - and speaks from actual deployment experience with real scale numbers, not pure thought-leadership; however, the episode functions partly as a vendor sales pitch for his current company, which limits candour.

“we've seen millions of interactions. Um, I think we're just shy of a million plus calls. Um, over 130,000, um, service booked appointments”

“we don't use third party tools like you know, um, N8 ends of the world or makes of the World to stitch workflows. That's actually one of our biggest, um, secret sauces. Took us about two years to develop”

The unnamed OEM tier-one case study (40% re-engagement at 6-7 weeks, 7-8 week purchase lag, liquidity-constrained buyer signal) and the 130,000 booked service appointments are concrete, but the OEM is deliberately unnamed, most ROI claims are illustrative rather than verified, and dealer-level metrics are presented as hypothetical ranges rather than actuals.

“The agent gave 30 minutes to the dealer for the first 30 minutes to respond to the lead...about 60 to 70% of this audience that would just stop...About six to seven weeks we saw 40% of this audience re engage”

“Over 130,000, um, service booked appointments”

Brian sets up a genuinely useful framing around consent-cadence-context and uses the Urban Science defection data to add substance, but he consistently affirms rather than challenges vendor claims, never pushes on pricing, failure rates, or competitive differentiation, and frequently finishes Albert's sentences rather than probing deeper.

“I love that, I love that”

“Come on, let's go. I love that. Albert”