ABOUT THIS EPISODE
A lot of candidates have gotten very good at saying the right things about AI in interviews. The problem is that saying the right words and saying them in the right order are two very different skills, and most hiring teams cannot tell the difference until it is too late.
Tom Andrews is VP of GTM and Revenue Operations at Hivebrite and principal at TA Advisory. He has taken a rev ops and enablement team from ten people to two without missing the output, and he has equally strong opinions about why most companies are trying to layer AI onto a data foundation that will never deliver real ROI. Tom has spent his career building the systems and teams that make organizations actually work, and on this episode he gets specific about what that looks like in an AI-driven world.
This episode is for founders and revenue leaders hiring for rev ops and enablement roles, anyone trying to assess genuine AI fluency in a candidate, and leaders trying to figure out whether to fix or rebuild a broken tech stack. Tom covers how to design interview tasks that actually filter out AI-assisted bluffing, why data architecture has to come before any AI investment, and why he believes most in-house rev ops teams are heading toward a leaner, agency-supported model.
Key Takeaways
- Anyone can say the right words about AI. The skill to look for is whether they say them in the right order.
- A well-formatted slide deck rarely comes from an LLM.
- The real skill of a modern leader is asking great questions, not generating long documents. A poorly contextualized prompt produces a generic report. A precisely framed one, with real business context, produces something genuinely useful.
- Most companies build a Frankenstein system: one tool bolted onto another, with no central data architecture. Fixing it is often more expensive than starting over. Choose your core platform, consolidate around it, and hire someone certified in that system who can create value from day one.
- Token efficiency is becoming a real cost center. A well-structured org with clean markdown files and a clear context layer can get the same output from a fraction of the tokens that a messy, siloed system requires.
- Bring in rev ops expertise earlier than feels necessary. The companies that build the scaffolding first avoid building a Leaning Tower of Pisa they will need to tear down and rebuild later.
Chapter Markers
(00:00) Cold open: why AI gets complex processes that humans struggle to describe
(01:52) How Tom took his rev ops and enablement team from 10 to 2
(05:05) Why in-house rev ops is becoming harder to justify
(09:21) The hidden cost problem: token usage and clean data
(13:29) Tool fatigue and the challenge of leading through constant change
(17:29) Spotting candidates who say the right words in the wrong order
(18:37) Designing interview tasks AI cannot easily pass
(22:09) Why hiring is one of the few things AI still cannot do for you
(26:38) Fixing versus rebuilding a broken tech stack
(31:22) Why Tom is going back to university to study machine learning
(31:59) The Frankenstein's monster system and why it happens
(35:19) Managing the cultural change to fix it for good
(43:01) Wrap-up
Useful Links & Resources
- Tom Andrews on LinkedIn: https://uk.linkedin.com/in/tommandrews
- HiveBrite: https://www.hivebrite.io
Connect With the Show
- Captivate Talent on LinkedIn: https://www.linkedin.com/company/captivate-talent/
- Danielle Parker on LinkedIn: https://www.linkedin.com/in/daniellemessler/
