Vetting the VCs: Avoiding 'Pilot Purgatory' Partners

Transcript

SPEAKER_1: supply chain AI is a painkiller, not a vitamin. Now I want to get into something that trips founders up after they nail that pitch — picking the wrong investor. SPEAKER_2: Yes, and it's quietly painful. A founder can have a genuinely strong product and still end up stuck — not because the market rejected them, but because their lead investor didn't understand what they'd signed up for. SPEAKER_1: So what does 'stuck' actually look like here? SPEAKER_2: Think of it as pilot purgatory. An enterprise AI project launches, generates promising early data, and then just... stays there. It never converts into a production system. The startup keeps servicing the account, and no one pulls the trigger on real deployment. SPEAKER_1: And a generalist VC doesn't recognize that pattern as a warning sign? SPEAKER_2: Exactly. They're benchmarking against SaaS timelines — pilots closing in weeks, revenue scaling fast. But supply chain enterprise integrations are a different animal. Sales cycles are longer, procurement is slower, and the technical lift is real. SPEAKER_1: What makes the technical lift so heavy? From the outside it looks like connecting software to software. SPEAKER_2: The key idea is that supply chain AI doesn't sit in isolation. It has to connect securely into ERP, WMS, TMS, and internal APIs. Each integration has its own security requirements and organizational gatekeepers. A rushed implementation often produces a brittle wrapper — works in the demo, breaks when a workflow changes. SPEAKER_1: So for someone evaluating a VC, what are the actual signals that an investor understands this world? SPEAKER_2: Three things. Do they ask about integration depth — secure connections into ERP, WMS, TMS, and internal APIs — rather than treating it as just a model question? Second, do they understand multi-tenant SaaS constraints, like tenant data isolation and churn analytics? Third, and this is the one most people miss: do they ask about post-deployment monitoring? SPEAKER_1: Why is post-deployment monitoring the tell? SPEAKER_2: Because AI systems drift. Data changes, models change, workflows change. A savvy investor knows shipping the product is the starting line, not the finish line. If a VC never asks how a team monitors agent behavior after launch, they're thinking about software — not living systems. SPEAKER_1: Now, there's another failure mode I've heard about — where the problem isn't the AI at all, it's the process underneath it. SPEAKER_2: This one is genuinely underappreciated. A surprising failure pattern in enterprise AI isn't model weakness — it's lack of operational context. For example, if a company automates a broken procurement process, they don't fix it. They just produce bad results faster. A strong investor should ask whether the process was mapped from real execution data, not just the happy path. SPEAKER_1: So due diligence goes both ways — founders should vet investors on operational design, not just market size. SPEAKER_2: Precisely. That includes asking whether they value human-in-the-loop design. The best supply chain AI deployments combine automation with oversight. An investor who pushes for full automation at all costs will create pressure that breaks trust with enterprise clients. SPEAKER_1: What about security and compliance? That feels like a place generalist investors underestimate. SPEAKER_2: It's a major gap. Enterprise clients will require audit trails, access controls, and frameworks like SOC 2 before signing anything meaningful. If a VC hasn't seen that diligence process before, they'll underestimate how long compliance readiness takes — and get impatient at exactly the wrong moment. SPEAKER_1: The takeaway for founders is really about matching investor patience to the actual integration timeline of this sector. SPEAKER_2: The goal isn't finding the investor most excited about AI. It's finding one who's already lived through a long enterprise integration cycle and didn't panic. Ask for live references from their portfolio — not polished demos, but companies actually running in production. That question alone tells a founder everything they need to know.