Interview: Zoom Dustin Bliss X Adam Demren — 2026-04-28
Key Themes
Cold start / data flywheel problem is the central strategic question. Both Bliss and Dustin framed the core challenge correctly: how do you get companies to contribute proprietary supplier data, and what service do you offer in exchange? Compliance automation (UFLPA, Chinese tech screening for government contracts) is the leading wedge candidate. The insurance/MGA model (Coalition Inc. analog) is the acknowledged long-term monetization layer, but everyone agreed you need the data asset first.
Compliance as a need-to-fix vs. nice-to-fix. Demren pushed hard on this distinction and it’s the right pressure. The team doesn’t have a clean answer yet — they have conversations scheduled in the next 10 days with chief compliance officers and CLOs to pressure-test this. The Cornell/DOE mapping project anecdote is a useful early signal: ops teams wanted it, legal teams killed it over NDA reconciliation. This suggests the compliance workflow is real pain, but internal gatekeepers are an obstacle.
Legal teams as the actual bottleneck. The Cornell lab example is a notable early data point — federal-funded semiconductor supply chain mapping was blocked not by ops but by lawyers citing NDA thickets. This implies the compliance wedge may need to meet legal/compliance teams where they are rather than routing around them.
Customer segment TBD. Fabless, device manufacturers, foundries — the team explicitly flagged this as an open question. Demren noted fabless semis are doing well financially, which may reduce urgency. The more interesting segment may be downstream device manufacturers or system integrators experiencing acute supply chain shocks (e.g., Skydio China sanctions, Base Power tariff impacts).
Shift Technology as a comparable. Adam flagged Shift (French AI fraud detection → claims automation, now $200M ARR) as a federation model analog — but noted it took 5–6 years to achieve data federation with insurers. This is a realistic benchmark for the long arc.
Notable Quotes
- Demren: “If the problem is more of a nice-to-fix versus a need-to-fix, you’re going to lose a lot of steam and momentum with a go-to-market.”
- Bliss: “The ops teams were pushing for it. The legal teams are pushing against it.”
- Adam: “There’s not that many firms you get to work with that are responsible for setting up their own firms.”
Surprises
- The Cornell/DOE mapping project being killed by NDAs — not by lack of interest or technical infeasibility — is a specific, actionable finding. Compliance teams are the actual gatekeeper.
- Adam’s portfolio company Base Power is experiencing exactly the kind of supply chain shock (promised 1,000 units/month, delivered 300) that the team should be targeting.
- The cybersecurity angle (semiconductor provenance in data centers, tracing upstream chip vulnerabilities) came up briefly but wasn’t developed — could be a distinct wedge worth exploring.
Open Questions
- Is UFLPA compliance a need-to-fix or nice-to-fix for target customers? What’s the penalty severity / enforcement track record?
- Which customer segment (fabless vs. device OEM vs. foundry) has the most acute near-term compliance pain?
- Can the legal NDA thicket problem be solved by offering compliance automation as the incentive to get lawyers to say yes?
- How long did Shift Technology’s federation journey actually take, and what were the unlock moments? (Adam offered intro to founder.)
- What is James at Substrate working on and why is timing TBD?