The businesses outperforming their competitors, according to Dr. Connor Robertson, founder of Elixir Consulting Group and host of The Prospecting Show, are not using more AI tools than everyone else. They are using them in a more disciplined sequence. Robertson lays out a five-layer AI stack he describes as the framework separating reactive AI users from businesses where AI is genuinely doing the operational work.
Layer 1: Intelligence Capture
Before any AI can work, Robertson says a business needs to be listening. Intelligent capture means every customer interaction, operational metric, and market signal is being logged somewhere structured. Most businesses capture some of this by accident, he argues; the goal is to capture all of it on purpose. Without deliberate capture, an AI system is being fed partial information, which produces partial results regardless of how powerful the underlying model is.
Layer 2: Knowledge Management
Raw data is not useful until it is organized, in Robertson’s framing. This layer turns captured information into a searchable, structured knowledge base an AI can draw from, the point where tools like Notion AI or a well-configured shared drive become load-bearing infrastructure rather than filing cabinets. His test for any document: can the AI find this when it needs it? If not, the knowledge management layer is broken.
Layer 3: Decision Automation
With a knowledge base in place, Robertson says a business can start automating recurring decisions: routing, scoring, prioritizing, flagging. He describes these as the decisions that eat roughly 30 percent of a team’s time and produce inconsistent results because they depend on whoever happens to be available that day. Decision automation, in his view, encodes existing judgment into a system that applies it consistently, at scale, without requiring anyone to be present.
Layer 4: Content and Communication Generation
This is the layer most people jump to first, according to Robertson, and it is the least powerful in isolation. AI writing and email drafting only compound in value when layers one through three are feeding them context. Without a knowledge base and decision logic upstream, he says AI-generated content reads as generic; with them, it becomes far closer to content written by someone who knows the client deeply.
Layer 5: Feedback and Self-Improvement
The final layer is the most overlooked, in Robertson’s account. An AI stack should be learning from its own outputs: did the email convert, did the proposal close, did the lead score predict actual behavior? Feeding that signal back into the system, he argues, turns a static automation into a compounding capability; without a feedback layer, a stack stays as good as it was on day one; with one, it improves over time.
Why the Order Matters More Than Any Single Layer
Robertson’s central argument is that most businesses do not fail at AI because they picked the wrong tools; they fail because they built the layers out of order. Jumping straight to content generation without a knowledge base underneath it produces output that sounds plausible but knows nothing specific about the business it’s supposed to represent. Automating decisions before intelligence capture is reliable means encoding bad judgment into something that now runs at scale instead of a human occasionally getting it wrong. In his framing, the five layers are not five separate initiatives to pursue in parallel; they are a dependency chain, and skipping ahead is the most common reason a stack underperforms relative to the tools inside it.
About Dr. Connor Robertson
Dr. Connor Robertson is an entrepreneur, author, and strategic advisor based in Pittsburgh. He is the founder of Elixir Consulting Group, host of The Prospecting Show, publisher of The Pittsburgh Wire, and founder of The Grant Finder. He is also a six-time published author, with titles including Built to Run, available at drconnorrobertsonbooks.com. More on his work is available at drconnorrobertson.com.
