The Refrigerator
Ross Sylvester, Co-Founder & CEO, Adrata | Mar 2026 | ~14 min read
In February 2023, Chamath Palihapitiya said something that lodged in my brain and never left:
"The people who invented refrigeration made some money, but most of the money was made by Coca-Cola which used refrigeration to build an empire. I view these large language models as refrigeration. Will there be some money made there? I think so. But the Coca-Cola has yet to be built."
It's a brilliant analogy. Refrigeration was transformative infrastructure. It changed everything about how food was stored, transported, and consumed. But the people who built the compressors and cooling units? They did fine. The people who understood what refrigeration made possible — the ability to distribute a sugar water drink to every corner of the Earth at a consistent temperature — they built one of the most valuable companies in human history.
Chamath's point was that in the AI era, the LLM providers — OpenAI, Anthropic, Google — are the refrigerator manufacturers. They're building crucial infrastructure. Some will do very well. But the real generational wealth will be created by companies that use AI to transform an industry so fundamentally that the before and after are unrecognizable.
I think about this every day. And I think he's right about the analogy. But I think he's wrong that the Coca-Cola hasn't been built yet. I think we built it. I think we're building it right now. And I think the reason people haven't noticed yet is the same reason nobody noticed Coca-Cola in 1886 — it looked like a pharmacy experiment, not an empire.
The Pharmacy in Atlanta
John Stith Pemberton was a pharmacist in Atlanta. In May 1886, he mixed coca leaf extract and kola nut with carbonated water in a brass kettle in his backyard. He sold it at Jacob's Pharmacy for five cents a glass.
Nobody thought this was the beginning of a $280 billion company. It was a tonic. A curiosity. The market for carbonated beverages was tiny, fragmented, and considered unserious by anyone in real industry.
Three things changed everything:
First, refrigeration. Before mechanical refrigeration became widespread in the 1890s and early 1900s, cold drinks were a luxury. Ice was harvested from lakes in winter and stored in insulated houses. Distribution was geographically constrained. Refrigeration didn't make Coca-Cola taste better. It made Coca-Cola possible everywhere. Suddenly, a cold Coke wasn't a seasonal indulgence in Atlanta — it was available in every diner, every gas station, every country on Earth.
Second, the bottle. Asa Candler, and later Robert Woodruff, didn't just sell a drink. They built a distribution system. The franchise bottling model meant Coca-Cola could scale without building every factory. They created a standard — the contour bottle — that became the most recognized object in commercial history. The system was more important than the syrup.
Third, the experience. Coca-Cola wasn't selling sugar water. They were selling a feeling. Refreshment. Happiness. Togetherness. The product was functional. The brand was emotional. The combination was unassailable.
Infrastructure. System. Experience. That's the formula.
The Revenue Organization Is the Pharmacy
Here is what I see when I look at the revenue organization in 2026:
Every CRO sits in a pharmacy. They have a hundred tools on the shelf. A CRM that was designed in 2004. A sales engagement platform that sends emails. A conversation intelligence tool that records calls. An intent data vendor. A forecast spreadsheet. An enablement platform no one logs into. A data enrichment provider. Another data enrichment provider. A third data enrichment provider.
They have spent $15,000 to $30,000 per rep per year on this stack. The average enterprise revenue team uses 12 to 15 tools. The integrations are fragile. The data is duplicated, stale, or contradictory. The reps spend 66% of their time on activities that don't directly generate revenue. That number comes from Salesforce's own research, and it hasn't meaningfully improved in a decade.
This is the pharmacy before refrigeration. Lots of ingredients. No Coca-Cola.
Now add AI. Every one of those vendors has added a chatbot, an AI copilot, an AI assistant, an AI-powered feature. The CRM has an AI. The engagement platform has an AI. The conversation tool has an AI. Every single one of them is doing the same thing: taking their existing product, wrapping a language model around it, and calling it innovation.
That is not Coca-Cola. That is putting a sticker on the refrigerator.
The Coca-Cola Is the System
What Pemberton and his successors understood — what Woodruff perfected — was that the value wasn't in any single ingredient. It wasn't the cola nut. It wasn't the sugar. It wasn't even the refrigeration. The value was in the system: a standardized product, distributed through a global bottling network, marketed with emotional precision, consumed as a ritual rather than a commodity.
The Coca-Cola of revenue is the same thing. Not a better CRM. Not a better email tool. Not a better dialer. Not another point solution with an AI wrapper.
The Coca-Cola of revenue is a system that orchestrates the entire revenue operation — from the first signal of buyer intent, through the multi-threaded complexity of an enterprise deal, across every handoff and stakeholder and objection and timeline — in a way that is intelligent, continuous, and unified.
That's what we built.
What the Refrigerator Made Possible
Let me explain what AI — the refrigerator — makes possible that wasn't possible before. Not theoretically. Actually. In our system. Running. Today.
Buyer group intelligence at scale. Before AI, mapping a buying committee was manual work. A rep had to research LinkedIn, ask their champion, guess at org charts. The median enterprise deal has 6 to 10 decision makers. Most reps know 2. Our system uses AI to identify, enrich, and map buyer groups automatically — who has authority, who has influence, who is the economic buyer, who is the technical evaluator. This isn't a feature. It's a fundamental capability that changes win rates.
Composition that adapts to the human. Our Autonomous Composition Engine doesn't generate generic emails. It builds prospect orientations — understanding the behavioral archetype of each person, their engagement patterns, their company signals, their competitive landscape. It composes emails that are specific to the individual, calibrated to their relationship warmth, optimized against seven dimensions of cognitive curiosity. A cold email to a methodical CFO reads entirely differently from a warm follow-up to a mobilizer VP. The system knows the difference. It adapts.
Orchestration that thinks. The revenue process isn't linear. Deals don't move through a pipeline in a straight line. They oscillate. They stall. They regress. They require different actions at different moments for different stakeholders. Our system watches the signals — engagement patterns, response sentiment, deal velocity, competitive indicators — and surfaces what matters. Not everything. What matters.
Forecasting that doesn't lie. Every CRO knows the forecast is wrong. It's based on rep self-reporting, pipeline stage, and hope. Our approach uses buyer group coverage, engagement depth, and signal triangulation to predict outcomes. Not what the rep thinks will happen. What the data suggests will happen. This is what AI enables that spreadsheets never could.
None of these capabilities exist in any single tool on the market. They exist because we built a system, not a feature.
The Dell Parallel
There's another analogy that matters. In the 1990s, Michael Dell didn't invent any component of a personal computer. He didn't invent the processor, the hard drive, the display, the memory, or the operating system. What he invented was a system for assembling, configuring, and delivering those components directly to customers faster and cheaper than anyone else.
Dell's insight was that the components were commoditizing. Intel made the chips. Microsoft made the OS. Seagate made the drives. The value wasn't in making those components. The value was in orchestrating them into exactly what each customer needed, at the right price, delivered in the right timeframe.
We see AI the same way. The models are commoditizing. Claude, GPT, Gemini — they're all excellent. They'll get better. The price will keep dropping. The capability will keep increasing. Competing on model quality is like competing on chip speed. It's a Red Queen's Race where you run as fast as you can to stay in place.
The value — the durable, defensible, compounding value — is in the system that orchestrates those models in the service of a specific, deep, complex domain. Revenue.
We don't compete with Anthropic. We don't compete with OpenAI. We use them. All of them. We orchestrate their capabilities across a hundred different tasks in the revenue workflow — composition, research, analysis, scoring, enrichment, forecasting, coaching — in a way that is so tightly integrated with the domain that the AI disappears. The user doesn't see a chatbot. They see their pipeline. They see their deals. They see their buyers. The intelligence is invisible, ambient, and continuous.
That's the Dell model applied to AI. Commoditized components, world-class orchestration, delivered as a transformative experience.
Why Nobody Notices
In 1886, if you walked into Jacob's Pharmacy and saw someone drinking Coca-Cola, you would have thought: another tonic. There were dozens. Patent medicines, herbal remedies, carbonated elixirs. The shelf was crowded.
In 2026, if you look at the revenue technology market, you see the same crowded shelf. Another AI sales tool. Another revenue platform. Another copilot. Another agent. The market is so noisy, so saturated with AI-washed marketing, that the signal is buried in noise.
But here's what's different about us:
We're a system, not a tool. When a rep logs in, they don't switch between 12 products. They work in one environment where every capability — pipeline, deals, contacts, emails, calls, analytics, forecasting, coaching, enablement — is integrated, intelligent, and connected.
We built the infrastructure. Rust and Axum on the backend. Not because it was trendy, but because processing millions of signals across thousands of buyer groups in real-time requires performance that interpreted languages can't deliver. Our API responds in single-digit milliseconds. Our system handles the computational load of real orchestration.
We're obsessed with the CRO. Not the rep. Not the SDR. Not the marketing team. The CRO. The person responsible for the entire revenue engine. We spoke to CROs. We studied CROs. We listen to CROs on podcasts while we prospect. We build for the person who needs to see the whole board, not a single piece.
We compound. Every deal we help close, every buyer group we map, every sequence we optimize, every subject line we score — the system gets smarter. The data compounds. The intelligence deepens. A new customer on day one gets the benefit of every insight we've accumulated from every other customer. This is the network effect that Coca-Cola built through distribution. We're building it through intelligence.
The Constraint That Became the Advantage
I learned something recently about constraints. I've had a constraint in my life — an invisible one — that I didn't fully understand until I was sitting in Mexico City trying to send emails. There was something stopping me from selling at full capacity. Not fear. Not lack of confidence. An old, deep pattern from childhood that triggered an alarm whenever I tried to sell.
When I surfaced it, it vanished. And what I found underneath it was the realization that the thing that held me back was the same thing that made me extraordinary at understanding this problem. The constraint and the gift were the same thing.
Adrata is the same. We were a small team building something ambitious in a market dominated by billion-dollar incumbents. That was the constraint. But it forced us to be faster, smarter, more integrated. It forced us to build a real system instead of bolting features onto a legacy architecture. It forced us to think like a CRO instead of thinking like an engineer. The constraint became the advantage.
The Formula
Coca-Cola's formula is infrastructure + system + experience.
Ours is the same:
Infrastructure: AI as the refrigerator. LLMs, embeddings, real-time processing, multi-agent orchestration. We didn't build the refrigerator. We use every refrigerator on the market.
System: Revenue orchestration. Pipeline intelligence. Buyer group mapping. Autonomous composition. Deal authority scoring. Sequence optimization. Forecast calibration. Coaching intelligence. All unified. All connected. All intelligent.
Experience: A CRO logs in and sees their world — not a dashboard, not a spreadsheet, not a report. Their world. The deals that matter. The people who decide. The actions that drive outcomes. It's not a tool they use. It's a system they live in.
Infrastructure. System. Experience. That's the formula. That's the Coca-Cola.
The Infinite Game
Chamath was right about one thing. This is an infinite game. The models will get better. The competition will get fiercer. New entrants will emerge. Categories will shift.
But Coca-Cola has been selling the same basic product for 140 years. The formula barely changed. What changed was the system, the distribution, the experience, the brand, the emotional connection. The product was always good enough. The system was always getting better.
We're at the beginning of this. We have the formula. We have the system. We have the experience. And every day we ship, every CRO we serve, every deal we help close — the moat gets deeper.
The refrigerator is amazing. But nobody remembers who built the refrigerator.
They remember Coca-Cola.
If your revenue team is still drinking from the pharmacy counter, we should talk. The Coca-Cola is here.