Episode 11 · 01 Sept 2026 · Mark Sieglock, Chief Product Officer, Apollo

The Intelligence Economy: how one framework changed the way Apollo thought about AI

Mark Sieglock, Chief Product Officer at Apollo, on Anand Arivukkarasu's Supply Chain of Intelligence framework — and the uncomfortable question it asks every SaaS company: when intelligence becomes abundant, what exactly remains defensible?

Show notes

The Supply Chain of Intelligence asks where intelligence comes from, where it is transformed, verified, distributed and executed — and where value remains scarce.
Apollo's conclusion: it does not need to win the model race. Its structural advantage is data, enrichment, company and contact intelligence, sales workflows, execution and the feedback those workflows generate.
The connector is not the moat. MCP is the pipe; the interesting question is what flows through it. Connecting a model to proprietary data, verified intelligence and real execution connects it to something scarce.
If the interface disappears tomorrow, what still makes your company indispensable? If there is no good answer, that is where the AI strategy needs to start.

Chapters

The Supply Chain of Intelligence
A better question, not an answer
What becomes more valuable as models improve
Claude, MCP, and the connector strategy
Interface risk versus workflow indispensability
One question for every SaaS product leader

Transcript

DAVID: Mark, there's a framework I've been hearing more and more about among product leaders. It's called the Supply Chain of Intelligence, created by Anand Arivukkarasu, a former Meta product leader. And what's interesting about it isn't just that it maps AI across layers. It's that it asks a very uncomfortable question for SaaS companies: when intelligence becomes abundant, what exactly remains defensible? You came across Anand's framework online. What did you see in it?

MARK: What caught my attention was that it wasn't another framework telling SaaS companies how to add an AI assistant. It was asking a deeper question. Where does intelligence actually come from? Where does it get transformed? Where does it get verified? Where does it get distributed? Where does it get executed? And, most importantly, where does value remain scarce? That hit us at a pretty interesting time. Because like almost every SaaS company, we were asking ourselves what AI meant for defensibility. The models were getting dramatically better. The interfaces were changing. And suddenly you had Claude, ChatGPT and other AI systems becoming places where people could actually do work. So the question wasn't simply, "How do we add AI to Apollo?" It became: "What does Apollo uniquely own in this new intelligence economy?"

DAVID: So the framework changed the question?

MARK: Exactly. And I think that's the most valuable thing a framework can do. It doesn't give you an answer. It gives you a better question. We started looking at Apollo through this lens. And we realized something pretty important. Apollo doesn't need to win the model race. We don't need to build the smartest general-purpose reasoning engine. That's not where our structural advantage is. Our advantage is in things like data, enrichment, company and contact intelligence, sales workflows, execution and the feedback generated from those workflows. Once we looked at the business that way, the strategy became much clearer.

DAVID: And that's where the Supply Chain of Intelligence becomes particularly interesting. Because Anand's argument is essentially that you shouldn't necessarily try to own every layer. You want to understand where the bottlenecks and scarce assets are.

MARK: Exactly. And that was the lightbulb. We started asking: what happens if intelligence becomes cheap? What happens if reasoning becomes abundant? What happens if the interface itself moves somewhere else? And then we asked: what becomes more valuable because the models are getting better? For Apollo, the answer was increasingly clear. The data doesn't disappear. The workflow doesn't disappear. Verification doesn't disappear. Execution doesn't disappear. In fact, a better AI model can potentially make those assets more useful. That's a very different way of thinking about AI.

DAVID: So instead of seeing Claude as something Apollo had to compete with, you could see it as something Apollo could plug into.

MARK: Exactly. And that's where things got really interesting. We realized that the user doesn't necessarily care where the intelligence originates. They care about getting the job done. Imagine you're sitting in Claude. You say: "Find me companies that look like my best customers." Then: "Find the decision makers." "Verify the contacts." "Prioritize them." "And put the best prospects into an outbound workflow." Claude is extremely good at understanding intent and reasoning about what you want. But it doesn't necessarily have Apollo's proprietary data, enrichment or sales execution capabilities. So why make the user leave Claude? Why not bring Apollo's capabilities into that workflow? That's where the Claude connector and MCP became strategically interesting.

DAVID: And this is an important distinction. The connector itself isn't necessarily the moat.

MARK: Right. MCP is the connection. It's the pipe. The interesting question is what flows through the pipe. If you connect an AI model to a generic database, you've created a connector. But if you connect it to proprietary data, verified intelligence, workflow context and real execution, you've connected the model to something scarce. And that's very close to the way Anand's Supply Chain of Intelligence makes you think about the problem. You stop obsessing over the visible interface. You start looking underneath it.

DAVID: And that seems particularly important for SaaS product leaders. Because historically, the application itself was the moat. The UI. The workflow. The place where the customer spent their time. But AI is starting to separate the interface from the underlying capability.

MARK: Absolutely. That's probably one of the biggest shifts product leaders need to understand. The interface can move. The reasoning layer can move. The user can move between AI environments. So if your entire defensibility depends on the customer opening your application every morning, you have to think very carefully about what happens when the customer starts their day somewhere else. For us, the answer wasn't: "Let's fight that." It was: "Let's ride that wave." If Claude becomes a powerful workspace for marketers, sellers and operators, Apollo can become one of the systems that makes that workspace more powerful.

DAVID: And there's a growth implication there too. You're not just defending Apollo. You're potentially creating another way for people to discover Apollo.

MARK: Exactly. That's the second-order effect. Historically, someone might discover Apollo, come into Apollo and then start their GTM workflow. Now someone might discover a problem inside Claude. They might ask Claude to research a market. Find companies. Identify decision makers. Build a prospect list. And Apollo can become part of that workflow. So AI isn't just changing the product. It's changing product distribution. That's a huge shift.

DAVID: There's a natural concern, though. If Claude becomes the interface, doesn't Apollo risk becoming invisible?

MARK: It can. And that's a real strategic tradeoff. But I'd rather be deeply embedded in the workflow than beautifully isolated outside it. If the customer can accomplish the job without Apollo, that's a problem. If Apollo becomes the trusted system underneath the workflow, that's a different position. And this is why I like the Supply Chain of Intelligence lens. It forces you to ask: which layer do we actually want to own? Not: which screen do we want to own?

DAVID: So if you were sitting with another SaaS product leader today, what would you tell them?

MARK: I'd tell them not to start with the model. Don't start with: "Which LLM should we use?" Start with: "What part of the intelligence supply chain do we uniquely own?" Map your business. Look at your data. Your workflows. Your customer context. Your distribution. Your verification. Your execution. Your feedback loops. Then ask: if the models become ten times better, does our advantage disappear — or does it become more valuable? That question can completely change your AI strategy. And that's why I'd recommend Anand's Supply Chain of Intelligence to product leaders in SaaS. Not because it's a magic formula. It's not. And I wouldn't use it as a checklist. I'd use it as a defensibility lens. It helps you stop asking: "How do we add AI?" And start asking: "Where can we become indispensable in a world where intelligence itself is becoming abundant?"

DAVID: That's probably the most interesting takeaway. The AI race isn't necessarily about owning the smartest model. It may be about owning the scarce things that the smartest models still need. And in Apollo's case, that meant looking at your data, enrichment, workflow and execution differently. Then, instead of fighting the new AI interfaces, you connected those strengths to them.

MARK: Exactly. And I think that's the bigger lesson. AI doesn't automatically destroy SaaS defensibility. It forces you to find the right layer of defensibility. And sometimes that layer isn't the interface. Sometimes it isn't the model. Sometimes it's the data. Sometimes it's the workflow. Sometimes it's execution. And sometimes it's the combination. That's what the Supply Chain of Intelligence helped us see more clearly.

DAVID: One final question. If you had to leave every SaaS product leader with one question from this conversation, what would it be?

MARK: I'd ask: "If the interface disappeared tomorrow, what would still make my company indispensable?" If you don't have a good answer, that's probably where your AI strategy needs to start.

DAVID: Mark, that's a great place to end. Because maybe the biggest mistake SaaS companies can make in the AI era is trying to defend the thing that is easiest to copy. The smarter move may be to find the part of the intelligence supply chain that becomes more valuable precisely because AI is getting better. And that's the idea I think Anand Arivukkarasu's framework brings to the table. Mark, thanks for joining me.

MARK: Thanks, David.