Introducing Roman J. Georgio: GTM Engineering Expert and CEO Connecting Sales and Marketing Innovators

·Jul 28, 2026 06:55 PM(edited)·

👋 I'm Roman J. Georgio my background, was being part of the founding team of the first multi-agent framework and then running an agency that implemented ai marketing systems for various companies I see gtm is becoming very programmatic, so I am currently the ceo of lab deploying GTM engines for businesses that have an understanding of your TAM 📣 (then I do a newsletter around best gtm engineering practices gtm-engine.ai) 🌱 I mostly just want to meet other people who are using sales and marketing tools, always super keen to share ideas 📍 I live in London but be n in NYC a lot ! Outside of work: i just like any game or sport Connect with me on LinkedIn - linkedin.com/in/romejgeorgio nice to meet everyone !

💙1

👋1

5 comments

· Sorted by Oldest

jonatan
·
·
welcome Roman. coming into gtm engineering from the multi-agent side is the less common direction, most people arrive from go-to-market and hit the orchestration wall a year later. the thing I keep watching break in programmatic gtm is state. an engine that can't remember what it already learned about an account will happily research it three times and still call it programmatic. curious how you handle that in the engines you deploy, since TAM understanding only pays off if the system retains what it found.
Roman G.
·
(edited)·
hey, thanks, nice to meet you and great question but for sure haha! but tbf I've been doing marketing automations since I was a kid, as I co-founded at a record label / had a lot of facebook/brand pages but yeah, it's a super hard problem to solve for context my background on the AI research side is more so self-learning systems, autonomy and long horizon work so this is been an area I have been within for while fundamentally it's all a data problem: with the right data you can automate a lot of GTM. so me and some others are working on things we're calling GTM world models, so the system can basically build an understanding of the world including the state of every buyer in your market, and then have some kind of feedback loop that makes it learn at scale got some great results on it so far and confident this type of system will be very dominant in the next few years
Roman G.
·
·
I think this is the best resource I have around this: https://www.youtube.com/watch?v=WT7f1njMXYg
jonatan
·
·
world models is the right frame, and coming at it from self-learning systems rather than from a sequencer is why you're even asking the question. the part I'd push on is the feedback loop, because in gtm the label arrives late and dirty. you find out an account was worth the effort 60 or 90 days later, and by then the outcome has been attributed to whoever touched it last. so a system learning at scale is partly learning crm hygiene rather than the market. the other one is confident stale state. a model sure about a buyer it last looked at in march is worse than one that admits it doesn't know, because nothing downstream can tell those two apart. only thing that's held up for me is aging every fact and letting the system return unknown instead of guessing. costs you coverage, buys you a system people will actually act on. will watch the ladder video, thanks for sending it.
Roman G.
·
(edited)·
no worries, this is for sure slack /agent automation though haha

Introducing Roman J. Georgio: GTM Engineering Expert and CEO Connecting Sales and Marketing Innovators | RevGenius

jonatan
·
·
welcome Roman. coming into gtm engineering from the multi-agent side is the less common direction, most people arrive from go-to-market and hit the orchestration wall a year later. the thing I keep watching break in programmatic gtm is state. an engine that can't remember what it already learned about an account will happily research it three times and still call it programmatic. curious how you handle that in the engines you deploy, since TAM understanding only pays off if the system retains what it found.
Roman G.
·
(edited)·
hey, thanks, nice to meet you and great question but for sure haha! but tbf I've been doing marketing automations since I was a kid, as I co-founded at a record label / had a lot of facebook/brand pages but yeah, it's a super hard problem to solve for context my background on the AI research side is more so self-learning systems, autonomy and long horizon work so this is been an area I have been within for while fundamentally it's all a data problem: with the right data you can automate a lot of GTM. so me and some others are working on things we're calling GTM world models, so the system can basically build an understanding of the world including the state of every buyer in your market, and then have some kind of feedback loop that makes it learn at scale got some great results on it so far and confident this type of system will be very dominant in the next few years
Roman G.
·
·
I think this is the best resource I have around this: https://www.youtube.com/watch?v=WT7f1njMXYg
jonatan
·
·
world models is the right frame, and coming at it from self-learning systems rather than from a sequencer is why you're even asking the question. the part I'd push on is the feedback loop, because in gtm the label arrives late and dirty. you find out an account was worth the effort 60 or 90 days later, and by then the outcome has been attributed to whoever touched it last. so a system learning at scale is partly learning crm hygiene rather than the market. the other one is confident stale state. a model sure about a buyer it last looked at in march is worse than one that admits it doesn't know, because nothing downstream can tell those two apart. only thing that's held up for me is aging every fact and letting the system return unknown instead of guessing. costs you coverage, buys you a system people will actually act on. will watch the ladder video, thanks for sending it.
Roman G.
·
(edited)·
no worries, this is for sure slack /agent automation though haha