Shantanu Shekhar, VP of Revenue Operations at Personio, funded his GTM engineering team by cutting two BDR heads and redirecting that budget into builders. Twelve months later, AE productivity is up 30%, pre-call research time dropped from two hours to 15 minutes, and 80% of MQLs run through an AI inbound SDR. He tells Noah and Andy exactly how he got there.

What makes this episode worth your time is the operational specificity. Shantanu doesn't talk about AI strategy in the abstract. He walks through the four-pillar charter he built, the agents his team shipped, the ones that flopped on adoption, and the build-versus-buy calls that didn't go as planned. If you're trying to stand up a GTM engineering function or make the case for one, this is the closest thing to a playbook you'll find.

Topics discussed:

• Four-pillar GTM engineering charter: culture, process, data, and systems sequencing

  
  

• Redirecting BDR headcount to fund GTM engineers and how to make that case

  
  

• Why GTM engineering embedded in RevOps eliminates an entire layer of alignment friction

  
  

• Building an attribution agent on Gong transcripts so attribution becomes a prompt, not a tool

  
  

• Research agent that cut AE pre-call prep from two hours to 15 minutes, driving 30% ARR lift

  
  

• Capturing 80% of MQLs through an AI inbound SDR and expanding from chat to multimodal

  
  

• Post-sales reachability agent orchestrating Zendesk, email, and Outreach to surface churn risk and cross-sell signals

  
  

• Evolving from a center-of-excellence model to specialized GTM engineers by segment

  
  

• Why shipping without a feedback loop kills adoption, and how to build the transition cycle

  
  

• What Shantanu actually tests for when hiring GTM engineers, and why technical skill is just the floor

Listen to more episodes: 

Apple 

Spotify 

YouTube

Everett Berry's definition of GTM engineering is deceptively simple: remove the technical constraints that stop companies from growing as fast as possible. But in this episode, he unpacks what that actually requires in practice, and most senior GTM leaders will recognize immediately that the bottleneck almost never comes from a lack of ideas.

From how Canva monitors customer social feeds at scale to detect poor graphic design and route it into outbound plays, to how Clay itself rebuilt its entire events invite system from scratch over two to three months of painful iteration before it worked, this conversation goes deep on tactics, org design, and where the role is headed.

Topics discussed:

·       GTM engineering as a builder discipline, not an evolution of marketing ops

·       The three-layer implementation hierarchy: data quality, process automation, net new plays

·       Why centralization matters even when GTM engineers are embedded across functions

·       Rep ride-alongs as the primary method for finding plays worth scaling

·       Clay's infrastructure stack: Audiences on ClickHouse, Sequencer, Agents, and MCP connectors to ChatGPT and Claude

·       How PLG companies run self-serve to sales-led conversion plays, and why enterprise expansion still requires humans

·       The failure culture leadership must create before the first GTM engineer can succeed

·       Why vibe coders are often the wrong hire, and Clay's interview process for testing process thinking

·       GTM engineering as a small, permanent tiger team rather than a scaling headcount function

Listen to more episodes: 

Apple 

Spotify 

YouTube

Prabhav Jain, CEO of 11X, opens with something you rarely hear from a founder in this category: AI SDRs, as the industry has framed them, don't actually work. The problem isn't the technology. It's that everyone is pointed at the wrong question. This conversation gets into what the right questions are, and how 11X has built its entire go-to-market, product, and customer qualification process around answering them.

From a multi-factor customer qualification model that disqualifies CEOs and CROs as sponsors by design, to a two-week deployment process and a living "17 problems" document they hand to DIY skeptics, Prabhav shares the operational specifics that most founders keep internal.

Topics Discussed:

• Why "AI SDR" is the wrong frame, and what actually works instead

• Multi-factor customer qualification criteria that screens for GTM maturity, channel fit, and operational ownership before the sale

• Three internal questions every team must answer before any pilot has a chance of succeeding

• Outbound signal strategy: why commoditized signals fail and how to find ones that actually convert

• Using PLG and product usage data to trigger personalized cross-sell and upsell outreach at scale

• SMS and WhatsApp as pre-call warming channels to lift inbound connection rates

• The "17 problems" sheet Prabhav sends to companies considering building in-house

• How 11X runs a two-week deployment, including mailbox warming, CRM mapping, and why voice agents take slightly longer

• The case for collapsing the 50-tool GTM stack into a single agentic platform

• What rev ops looks like when sub-agents own execution and humans own optimization

James Underhill, Head of GTM Ops at Profound, runs the ops and systems function through one of the more extreme growth curves you'll hear about: five times ARR in nine months, headcount from sixty to two hundred, and an AE team scaling from thirty toward a hundred by year end. His operating principle is simple: buy infrastructure, build applications. That single bias shapes how his team is staffed, what they buy, what they build, and where they draw the line on both.

He built a deal desk bot in under twenty minutes on Dust without writing a line of code.  His team deflected 70% of support inquiries with an internal triage agent. And he replaced the traditional business partner role entirely by giving the field direct, semantic access to their own data.

Topics Discussed

• Why calibrating to the first derivative of growth matters more than current headcount

• Using Dust as a low-code agent platform across GTM, CS, and internal ops

• Deflecting 70% of support inquiries with a confidence-gated triage agent

• Buy infrastructure, build application as a disciplined team operating principle

• Why GTM engineering at this level requires actual software engineers, not just technical curiosity

• GitHub fluency as the new hiring proxy for this function

• Snowflake plus a semantic layer as the foundation for real-time, conversational data querying

• Centralizing Claude Code skills in Notion for rep-accessible, field-ready workflows

• The hidden maintenance cost of vibe-coded tools and why it compounds fast

• Replacing business partner roles with self-serve data access

• Where Gong's call intelligence falls short and why that becomes a build opportunity

• How to decide what to build versus buy when your team could technically do both

Seth Marrs spent six years at Forrester as their lead sales AI and technology analyst, which means he evaluated every major wave of GTM tech and has the receipts to back his calls. Now at Sandler as Chief Strategy Officer, he's seeing the practitioner side up close. His read on the agentic SDR market is blunt and sits well outside the vendor narrative, and he supports it with something most guests can't: actual data on what happens to sellers when you train them, test them, then watch them on live calls the next day.

He also gets specific on the parts most people gloss over. Why removing the seller from data capture is the first move that makes everything downstream possible. Why "next best action" is a flawed idea that quietly argues against needing sellers at all. And why the agentic opportunity in sales may sit with the SE and the human selling skill, not the SDR everyone is trying to automate away.

Topics discussed:

• Why "fire your BDRs and automate with AI" is the vendor red flag to watch for

• The recording-permission gap that breaks the data flywheel on outbound

• Keeping BDR research depth while moving from 10 calls to 50

• Sales motion as the deciding factor: high-velocity vs. long-cycle BDR economics

• Removing the seller from data capture entirely to build a reliable dataset

• Compete score: revenue per rep normalized across net-new, farmer, and CSM roles

• The 40 to 80 to 40 adoption collapse measured on live calls, not surveys

• Certifying sellers on what they actually say to customers, not test completion

• Why top-three suggested actions beat a single prescribed next best action

• The agentic SE: training a bot on the technical layer while humans own the conversation

• Extracting intelligence from the bottom of the funnel where capture doesn't exist

• Buy the infrastructure, build the AI layer on top: the maintenance trap with vibe-coded tools

Ryan Milligan started at Quotapath as Director of RevOps, spent four and a half years building the data and systems foundation, and is now CRO. His team has run at 100% blended quota attainment in 8 of the past 10 quarters, never below 90%, and has grown closed ARR per rep 1.7x in 18 months. The GTM engineering team doing most of the building is two people.

In this episode, Ryan gets specific about how the whole system works: the data architecture decision he made on day one that still underpins everything, how he splits Dust and Claude into distinct roles across the sales cycle, and why he thinks the current wave of everyone building their own tools is a bubble with a painful correction coming. He also makes a sharp case that comp plan design is one of the highest-leverage tools a CRO has for changing the mix of revenue being closed, not just paying people.

Topics discussed:

• Processing data in the warehouse nightly and reverse ETL-ing into CRM so both always speak the same language

• The build vs. buy litmus test: uniquely bespoke and relatively fixed vs. everything else

• Rep-built V1 prototypes handed to RevOps for productionizing and org-wide rollout

• Dust as system of record, Claude as system of action, and how they split across the sales cycle

• Thursday multi-thread standup: every rep required to arrive with all multi-threads queued for active opps

• The "what would it take to close twice as many deals" framework for identifying which agents to build next

• Warm outbound architecture using Clay, Unify, and product interaction data as intent signals

• Comp plan design as a lever for changing the shape of revenue reps close, not just incentivizing volume

• Why data architecture is the only real defense against confident AI hallucination

• GTM engineer defined as the owner of the full prospect-to-renewal lifecycle

Listen to more episodes: 

Apple 

Spotify 

YouTube

Mark Deacon, CRO of CaniBuild, has gone further than most leaders talking about agentic GTM. He's actually built it, measured it, and has the numbers to back it up: a 400% improvement in revenue per human headcount and a demo-to-close rate now sitting above 60%, more than double what it was before AI.

What separates this conversation is the operational depth. Mark walks through the exact sequencing logic behind their AI SDR workflow, the buy vs. build decision criteria he applies to every tool, how he onboards and governs AI agents the same way you would a new hire, and the centralized AI operating system he built from scratch to keep an 80-person company running with consistent governance across the stack.

Topics Discussed:

• 400% revenue per headcount improvement and 60%+ demo-to-close rate after AI deployment

• SMS-first sequencing strategy that increased AI SDR pickup rates through A/B testing

• ICP based routing logic that books demos directly into the right rep's calendar

• Buy vs. build decision framework based on uptime requirements and maintenance cost

• Two-to-three month AI agent onboarding process before handoff to the business owner

• Slack-native AI chief of staff architecture that routes tasks across a team of specialized agents

• One-script Claude Code config deployment for consistent governance across all team members

• AI-first vs. AI-only operating model and why the 80/20 split on support tickets matters

Listen to more episodes: 

Apple 

Spotify 

YouTube

When Scalestack audits a new enterprise prospect, CRM data quality typically comes back at 30-40%. That's the starting point for most companies trying to run AI agents across their GTM motion, and it's why most of those initiatives quietly fail. Elio Narciso, who left AWS to build Scalestack, makes the case that the missing piece isn't the AI layer, it's the orchestration middleware that sits between your data sources and your activation layer, and that without it, AI doesn't produce bad outputs, it weaponizes your existing bad data at scale.

What makes this conversation worth your time is that Elio goes well beyond "clean your data." He gets into the mechanics: why deciding when NOT to use AI and to use simple automation instead is one of the most important cost and scale decisions a GTM team can make, why dropping structured CRM picklists in favor of unstructured data may be one of the most underappreciated shifts happening right now, and why the GTM engineer role as it's currently defined is already becoming outdated, with software development as the more honest blueprint for where revenue teams are headed.

Topics Discussed:

• Enterprise CRM data quality averaging 30-40% at the point of AI deployment

• The orchestration middleware layer and why it couldn't exist before modern AI

• How forward deployment engineering translates business logic into agent missions

• The build vs. buy inflection point: when to stop experimenting with Clay and Claude and standardize

• Confidence scoring and agent reasoning trails as a replacement for data trust

• Why structured CRM picklists are becoming a liability as AI-driven data search replaces manual filtering

• Automation vs. AI agents: the cost and scalability decision most teams are getting wrong

• Why the GTM engineer title is already passé, and what software development tells us about what comes next

Listen to more episodes: 

Apple 

Spotify 

YouTube

Matt Millen spent years running revenue at Outreach, watching companies stall out post-onboarding, and built Regie.ai to fix the problems that sales engagement itself created. He was attaching generative AI to tools like Outreach and SalesLoft years before ChatGPT, which gives him a rare vantage point on what actually works vs. what's still hype.

In this episode, Noah and Andy get into why half of all AI POCs were failing (hint: it was both sides' fault), how Regie thinks about workflow before product, and why Matt believes "where humans enter the loop" is a brand decision, not a platform limitation. He also shares a blunt breakdown of which companies are good fits and which ones waste everyone's time, and makes the case for why the buy vs. build debate is being driven by CEOs who haven't thought through the workflow complexity underneath.

Topics Discussed:

• Why 50% of AI POCs were failing and the two-sided imbalance that caused it

• The four readiness signals Regie uses to qualify or disqualify a prospect before the POC starts

• Workflow interviews with frontline reps vs. managers and why the gap between them matters

• Shifting from day-17 sequence dumps to signal-triggered task lists with built-in call context

• Why human-in-the-loop placement is a brand decision, not a product constraint

• Seat-based pricing with bundled AI credits and how the 80/20 on data consumption actually works

• The workflow complexity case against building agentic sales in-house

• Where the SDR and AE roles are heading as agents absorb more of the top-of-funnel motion

Listen to more episodes: 

Apple 

Spotify 

YouTube