GTM Engineer: Gina @ Dust

Dust keeps coming up across the GTM engineering community, and this episode explains why. Gina Kabasakalis, Founding US GTM at Dust, has a clear-eyed view of where AI adoption actually breaks down: not in the build, but in the handoff from one technical person to an entire team.

She gets into the governance and permission infrastructure that makes multiplayer AI viable, with a concrete example of how RevOps encoding the right Salesforce schema field into a shared agent saves every sales manager from carrying that knowledge individually. She also covers what companies consistently get wrong about ROI measurement, and a new Dust feature called Pods that puts humans and agents in a shared workspace. If you're past the single-agent stage and trying to figure out how to scale this across an org, this one is worth your time.

Topics discussed:

• Why RevOps and IT/CISOs are the two buyer profiles gravitating toward Dust

• Multi-model routing: matching model complexity and cost to the task rather than defaulting to the highest-reasoning model for everything

• Encoding institutional knowledge at the system level so ICs stop carrying it individually

• Piloting vs. rolling out: validating long-term fit vs. the executive air cover and embedded AI ops function a real rollout requires

• Why time savings is the wrong AI ROI frame and what the CRM hygiene argument actually proves

• Pods: a shared workspace for humans and agents, with use cases from new hire onboarding to funding round comms

• Four-component agent framework: trigger event, data sources, recipient and destination, output transformation

• Agent sprawl: when to consolidate duplicate skills and when distributed creativity is the point