Revenue Managers were the Original ChatGPT — Photo by Infinito

Before you roll your eyes - yes YOU! Hear me out:

Everyone is obsessed with LLMs (Large Language Models) right now. ChatGPT, Claude, Gemini… all these super-genius tools that:

  • take in huge data sets
  • learn patterns
  • reason (sort of)
  • predict outcomes
  • patiently answer all the questions you have
  • answer the same question always differently :-)
  • and occasionally hallucinate with absolute confidence

Sound familiar?

Yeah. That’s literally revenue management… since forever.

1. LLMs learn from massive data.

Revenue Managers? We learned from garbage data. Incomplete data. Wrong data. Data that lived in 18 different spreadsheets created by someone who left in 2014.

LLMs get curated training sets. RMs get: a PMS, an RMS, some pace, a suspicious Excel file called Final_Final_V12_Actual, and a prayer.

And yet… we still manage to explain demand curves with more conviction than an AI explaining quantum physics.

2. LLMs find relationships between patterns.

This is basically the RM brain:

“Pickup is soft.” → Must be because: Event cancelled? Competitor moved? Weather? Guest sentiment? Some random flight schedule change? Or the universe hates Mondays?

It’s like mental prompt engineering.

Except ours is manual. And our output isn’t “generate a poem about cats.” It’s “Here’s a compression night, let’s squeeze another €27k out of it.”

3. LLMs hallucinate.

Let me be clear: ChatGPT didn’t invent hallucinations. Revenue managers did.

We’ve been saying “The model is confident” while being 100% unconfident since before ChatGPT was a twinkle in OpenAI’s eye.

Forecasts? We hallucinate optimism. Budgets? We hallucinate miracles. Owner meetings? We hallucinate a calm face while mentally screaming.

But at least our hallucinations usually come with a backup plan, three scenarios, and a recommendation slide.

4. LLMs improve based on feedback loops.

RMs? We improve based on trauma loops.

Owner didn’t like pace last month? → Adjust. GM questioned your strategy? → Reinforce. Sales did something random again? → Recalculate. Marketing launched a campaign without telling you? → Breathe. Strategize. Try again.

LLMs have reinforcement learning. Revenue Managers have scar tissue.

5. LLMs are used to answer complex questions quickly.

Which hotels used first? Revenue Managers.

“Why are we not full?” “Why are we too full?” “Why is ADR down?” “Why is ADR up?” “Why did pickup change?” “What will next month look like?” “What’s the meaning of life?” “Can we have the forecast by 2 PM?”

We were the original chat interface. Except our API limit was “one coffee per hour.”

6. And imagine two RMs talking to each other…

It’s basically two LLMs flirting:

RM1: “My comp set is misbehaving.” RM2: “Tell me more about your segmentation.” RM1: “Only if you show me your booking curve.” RM2: “Depends. How dynamic are you feeling today?”

Peak nerd romance.

The only difference? If two AIs flirt, they exchange embeddings. If two revenue managers flirt, they exchange trauma from last budget season.

The Real Point

AI won’t replace revenue managers, it’s simply becoming the upgraded version of the tools we always wished we had.

But the thinking? The commercial instincts? The ability to sniff demand before the data even hits the report? The human nuance of pricing psychology and upselling?

That’s RM territory. Always has been.

AI is the engine. Revenue Managers are the driver. And together? We’re unbeatable.

Love, Fabi

PS: I’m Fabian Bartnick — Founder, Commercial Growth Leader, AI-Powered Hospitality Innovator and Vibecoder. I’ve built and exited hospitality tech companies, angel-invested in next-gen startups, and trained thousands of leaders across 100+ countries. I help hotels and brands unify sales, marketing, revenue, and data into one unfair-advantage growth engine and yes, I still love a good revenue management debate.