The Frankenstein System Killing Your AI ROI with Tom Andrews

**Tom Andrews** (0:00)
The more complex the processes you're building, the harder they are to describe. And AI just gets it. We're going to see incredible levels of disruption over the next few years.
I think in-house RevOps is becoming harder to have a really good reason for, because the skills you need to maintain a modern tech stack are way more than any one person has.

**Danielle Parker** (0:30)
Welcome to Human-First, the GTM Hiring Show. I'm your host, Danielle Parker, Head of Marketing at Captivate Talent. There is a growing problem in how companies are hiring for AI fluency right now. A lot of people have got very good at saying the right things in interviews without having actually done anything. And for founders who are not themselves technical, that gap is almost impossible to spot without the right process. Our guest today has some very specific opinions about how to close it and some equally direct views on why most companies are trying to deploy AI on a data foundation that will just not work and will never get them the real ROI from it. Tom Andrews is VP of GTM and Revenue Operations at Hivebrite and Principal at TA Advisory.
He is someone who has spent his career building the systems and teams that make commercial organizations actually work. Welcome, Tom.
We caught up, I don't know what was it, like two weeks ago.
You mentioned that you've taken your RevOps enablement team from, I think it was a team of 10 down to two, and you do not miss the headcount. No. That's pretty bold statement.

**Tom Andrews** (1:42)
Yeah.

**Danielle Parker** (1:42)
So, what made that possible, and what does it tell us about what the roles were actually doing before and also how things are changing in RevOps right now?

**Tom Andrews** (1:52)
Great question. I think there's a couple of different factors of play. So, number one, a lot of teams are too big, and there's a lot of people doing things that they think are important, which to be really frank, aren't that important anymore. The world has changed quite a lot, and when you see really overbaked process, all it does is add friction and slow things down. The most important thing in modern business is efficiency, and that's why AI has caught on so quickly. In the company I work for now, and in a previous company, I've seen 70 to 80% headcount reduction within my team, but also in other teams as well. And unfortunately, it's just gonna be part of the future. There are a lot of people who follow kind of tried and tested traditional methods for getting things done, but there's a reason that the old school is the old school, and there's kind of this new establishment in place of people who get things done much faster. I was watching a release yesterday of a really exciting new product, and they were talking about how it used to take a team of five to do what can now be done just by one person. And so if you think about people in those classical VP roles, they know all the strategy, they've got all of the great ideas, or at least they should have, otherwise they shouldn't be VP's.

**Danielle Parker** (2:59)
Anyway, that's beside the point.

**Tom Andrews** (3:02)
The bit that's really interesting is they used to need a team of managers and ICs and specialists to get things done. Now, you can make a lot of that identical. I also think that when you see big transformations in RevOps and enablement especially, technology, regardless of whether it's AI or not, is a big part of the reason. The technology that's coming out can do things faster and better with zero errors. And honestly, in most cases, it's just a lot more effective. I recently took a 40 tab Google Sheet that is the kind of source of truth for all of the new pricing model that we've built. Loaded up into Gemini, asked Gemini to build me all of the enablement material we needed. I copied and pasted that into Confluence which we use for all our enablement and we were done. There's probably still things to tighten up and an epic queue to build as more questions come out. But before that would have taken people weeks to get through every single tab, digest it, understand it. I don't have to explain myself or explain what I've built to an AI.
That's crucial because the more complex the process is you're building, the harder they are to describe and AI just gets it. The other bit that's really impressive about most of the new generation of the LLMs, they can actually describe what they're seeing. So you can upload a video of you clicking through things. It can write the walkthrough. It can give step-by-step instructions. Again, that used to take hours. People would use complex technologies like Scribe to track all the clicks. Probably took a million screenshots in the process. Now, you just shoot a video.