Trying out a new podcast episode on the Crew podcast/blog this week!
The inspiration for this type of episode/article was from the feedback we got at the Agents in GTM event we threw at Clay HQ in March (RIP vibescaling).
At that event, Nic Siegel (founding AE that we placed @ Dust on the recruiting side) demo’d how he uses Dust to automate a bunch of mundane tasks in his day-to-day, and it really blew the audience away.
From there, we’ve been talking about more of a Greg Isenberg-style video - and was happy we could finally get in studio and do this.
Open to all feedback - let me know if we should be doing more (or less) of these types of one-off episodes.
Goes without saying, but this is best viewed on the YouTube channel for this.
Once again, appreciate our sponsors for this episode, Dust.
Dust: A shared workspace where GTM teams and AI agents coordinate complex work together instead of in silos. Think account command centers, RFP deal rooms, and cross-functional launches. Try them today @ https://dust.tt/
Links To Sections
Discussed In This Episode
The first live build episode on the show: Chris builds a working pre-call prep agent from scratch while Nic looks over his shoulder
Where companies actually are on AI maturity, and why so many pilots produce no ROI
The three stages of adoption: coding agents, then individuals on their laptops, then actual team systems
Why open source Chinese models matter, and the test where the same simple task cost 178 times more on one model than another
The MCP timeline, and the moment tools started talking to each other
The core sales use cases: an ask-sales bot in Slack, meeting prep before and after, prioritization and signals, call coaching off your own transcripts, and RFP and security questionnaires
How Clay and 1Password govern who is allowed to build what, and how good agents get promoted to the whole team
"Show me something you've built": the question Nic asks candidates, and the red flag when there is no answer
Describing an agent out loud with Whisper Flow instead of prompt engineering it by hand
Human in the loop controls set per tool action, not per agent
Why a cloud based system beats agent files sitting on one person's laptop
Pods: an isolated file system and sandbox where a team can build its own internal apps
Key Timestamps
(03:56) Nic's background, and what Dust actually is
(06:11) The three stages of AI adoption inside companies
(08:30) The MCP timeline, and when tools started talking to each other
(12:54) The core sales use cases, from ask-sales bots to RFPs
(18:06) How Clay and 1Password govern who builds what
(20:51) "Show me something you've built"
(22:04) Building a pre-call prep agent from scratch
(24:48) Picking a model, and the test that cost 178 times more
(39:03) Setting a 7am trigger and emailing yourself the brief
(49:26) Nic's cockpit: Pods and an AE operating system
Thanks for tuning in!
If you enjoy it, please give us a rating, review, or follow on Spotify/YouTube/Apple Podcasts - it really helps us grow this.
For those who are new, my name is Chris Balestras, co-founder & head of talent, media, & brand @ Crew - a GTM recruiting, media, and investing firm, working with seed through series D AI-natives to help them grow.
Where to find Crew:
We work with many of the hottest AI-native startups in various capacities, and for those who are interested, shoot me an email at ( [email protected] ) or a DM on LinkedIn.
🫡 cheers,
Chris