Interview: Zoom Dustin Ross Skiffra — 2026-05-01
Key Themes
Operational AI for Heavy Industry: Skiffra’s core thesis is that legacy enterprise software (SAP, ERPs) captures data but fails to deliver it in real time, at the right granularity, to the people actually making decisions on the floor. Their intervention layer sits between raw sensor data and the human decision-maker — not the executive suite.
Proof of Concept Credibility: The Indonesian mining POC is the anchor of the pitch. $7M in fees → $100M permanent annual EBITDA lift is a striking ratio. The truck-loading example is concrete and memorable: volumetric cameras + weight sensors + a red/green interface eliminated a $30M/year underloading problem. This is not vaporware — it’s deployed and measured.
“Hardware Is Hard” as a Moat: Andy explicitly frames physical-world complexity as a competitive barrier. The difficulty of instrumenting real environments, dealing with safety constraints, and embedding into operations creates defensibility that pure software plays lack.
Team Profile: Alex Smith (COO, Wall Street + mining turnaround background), George (Toyota digital transformation), Andy (VC + operating background). Early stage — 2 founders, ~10 international team members, 4 open engineering roles. Started formally early 2024, but POC goes back ~1.5 years.
Relevance to Project TBD: The overlap is indirect but real. Skiffra is building operational intelligence for physical supply chains (mining → metals → critical minerals). Their data layer — real-time, federated, decision-loop-oriented — is adjacent to the supply chain intelligence problem TBD is mapping. The semiconductor supply chain has critical minerals inputs; Skiffra is upstream of that. Worth monitoring as a potential data partner, channel, or reference model.
Notable Quotes
- “SAP is great at reporting, bad at predicting.”
- “We never think about what the guy operating the loader needs — we don’t think of him as a decision maker.”
- “Hardware is hard — that’s a real frontier now.”
- “You’re always going to need to get the rock through the processing and through the smelter. AI is not going to change that fact.”
Surprises
- The ROI ratio ($7M fees → $100M EBITDA) is unusually high and immediately credible because it’s grounded in a specific, measurable operational fix — not a vague efficiency claim.
- Andy came in with very little context on Project TBD, mirroring Dustin’s lack of context on Skiffra. The meeting was more mutual orientation than a targeted interview.
- Film industry was mentioned as an adjacent target — unexpected, but reflects their broader “data meets atoms” framing.
Open Questions
- What is the data ownership/licensing model for the operational data Skiffra accumulates across mining sites? Could this become a commodity intelligence asset?
- How does Skiffra’s federated data model interact with potential downstream users (insurers, traders, supply chain intelligence platforms)?
- Is there a direct connection point between Skiffra’s critical minerals visibility and TBD’s semiconductor supply chain thesis?
- What is the path from mining operations intelligence to a product that serves semiconductor supply chain stakeholders?