SaaS Interviews with CEOs, Startups, Founders

The episode delivers real operational data points (170% NRR, 2.5x YoY growth, land-and-expand trajectory from 5-7 to 100+ facilities) and a clear mechanistic explanation of the inventory error problem, but significant airtime is consumed by a YouTube fund-pitch interlude, background filler, and a mental health tangent unrelated to B2B operations.

“our net revenue retention annualized is 170% annualized”

“it starts with a little enterprise deal but expands into network wide fairly quickly and fairly often”

The hardware-agnostic angle (off-the-shelf Best Buy drones as data gatherers) and self-supervised barcode neural nets are genuinely interesting technical differentiators, but the broader framing follows a well-worn 'foundational model for physical AI' narrative and the ChatGPT-in-2015 analogy is already a common tech comparison template.

“our stack is the only stack in the world that you can put on a moving camera that you buy out of Best Buy and turns it into an autonomous data gatherer”

“physical AI is at the same point where ChatGPT was in around 2015, where it's just finding its roots”

Sankalp is a legitimate deep-tech operator - CMU PhD in robotics, DARPA autonomous helicopter, and a company with real enterprise revenue and elite retention metrics - not a thought-leader or career podcast guest; however, at ~$15M ARR he is still early-stage, limiting the scale of battle-tested lessons.

“the three co founders, Daniel Gitesh and myself, worked on the world's first fully autonomous helicopter, won national awards for the work we did with aerial autonomy”

“we grew 2.5x year over year last year”

The episode is notably concrete for its length: named customers (Barrett, Stadia Goods), a specific cost-per-pick comparison ($2 vs $10-15), a 70% error-reduction figure, average ACV ($500K), largest ACV ($3-4M), logo count (30-40), TAM framing (150,000 US warehouses over 100K sq ft), and first-code to first-customer timeline (2018-2021) all appear; depth of proof per claim is thin but breadth of numbers is above average.

“a pick that was supposed to cost $2 ends up costing $10 to $15”

“reduce those kind of location errors by 70%”

The host does extract specific numbers through direct follow-up questions and stumbles onto the forklift product via screen-sharing, but momentum is repeatedly broken by a lengthy fund-pitch ad read, simple arithmetic questions the guest could answer more usefully, and no substantive pushback on any claim (e.g., the 170% NRR or the TAM math goes completely unchallenged).

“I am not just a YouTuber. I'm investing in my third fund. We've deployed $250 million into 550 software companies”

“can I multiply the 30 logos times a half million bucks a year? That would put you at about a 10 or 15 million run rate”