Podcast
June 25, 2026
Episode 97: Everybody Has AI. Nobody Owns It.
Brandon Bussey on why giving everyone AI tools without governance creates agent graveyards — and how to build the hub-and-spoke model that actually works.
What happens when every team in the company gets access to AI tools at the same time, with no one coordinating how those tools get built, maintained, or retired? You end up with a problem that revenue operations has seen before — just with a different name on it.
Brandon Bussey, VP of Revenue Operations at Advantive, joins co-host Matthew Volm to talk through one of the most pressing organizational questions in RevOps right now: who actually owns AI? Brandon brings a grounded, finance-first perspective to the conversation — he grew up in FP&A, spent time at Concur Technologies and Amazon before pivoting into RevOps, and has since built revenue operations functions at Qualtrics and Lucid Software. At Advantive, a private equity roll-up spanning seventeen companies in manufacturing technology, he's dealing with AI governance challenges at a scale that makes the stakes concrete. This isn't a theoretical conversation about AI strategy — it's a working operator describing what's breaking and how he's trying to fix it.
Most companies made the same first move with AI: hand out licenses, let teams experiment, and see what happens. Brandon is quick to say that's not wrong as a starting point. But staying there creates a specific category of problem that RevOps practitioners will recognize immediately.
"I've got sales reps building these agents, they're connecting to the Salesforce MCP, pulling a bunch of data and guess what? It comes full circle to where now I've got a sales rep publishing some like HTML dashboard that they've created, and guess what? They didn't use the right filters based on our definition of bookings." — Brandon Bussey
The governance problem is one layer of it. But Brandon surfaces something that cuts deeper: even when the outputs are technically accurate, the question is whether building them is the right use of a sales rep's time in the first place. AI has made it easy for everyone to produce more — more dashboards, more agents, more automations. What it hasn't done is make those outputs more valuable or more aligned with what the business actually needs.
This connects to a pattern RevOps teams have lived through before. Matthew draws the parallel explicitly: the dashboard graveyard. Executives request views by channel, by segment, by rep. RevOps builds them. They get opened once and forgotten. The same dynamic is playing out now with AI agents — built quickly, used briefly, and then left running (or not running) without anyone tracking whether they're still producing quality output. As Brandon puts it, the agent graveyard.
The FP&A analogy is worth sitting with. A fully centralized analytics team can produce rigorous, consistent outputs — but without proximity to the business, they build things that don't fit how people actually work. The same risk applies to a centralized AI team. They have the technical capability, but not the context. Push too far in the direction of centralization and you end up with agents that get deployed and ignored.
The answer Brandon is building toward is something in between. He describes it as a hub-and-spoke model: a centralized team of agent builders at Advantive, with an intermediary layer sitting on the go-to-market operations team to manage the flow of requests and own what gets built.
"Someone that will sit on our kind of go-to-market or commercial operations team that will start to funnel requests, and then they can build or they can throw to the central team to build, but they are the gatekeepers. And then once something's built, they own the agents going forward." — Brandon Bussey
The job title: Go-to-Market AI Specialist, rolling up into revenue operations.
The ownership piece is what most companies have skipped entirely. Building an agent is easy. Deciding whether to build it, designing it to fit the actual workflow, monitoring it as the business changes, and retiring it when it's no longer useful — that's product management. And that's exactly the framing Brandon and Matthew converge on: the best operators in this environment are the ones who treat what they build as a product, with real end users and real accountability for outcomes.
This is a bigger shift than it might appear. Revenue operators are increasingly being asked to function less like administrators and more like architects — owning systems that have downstream effects on how work actually gets done. Layering AI into that responsibility doesn't change the underlying principle; it raises the stakes for getting the governance right.
One of the more practically useful things Brandon describes is a principle for where to actually deploy agents: build them inside the tools where the work is already happening, not in a separate centralized interface that people have to go find.
At Advantive, this shaped a deliberate tech stack evaluation. About a year ago, with several contracts coming up for renewal, Brandon assessed AI-forward alternatives to some of the tools his team had been using. The result was bringing in Glyphic (now Airspeed) as a call recorder with AI central to the user experience, and Letter AI for content management — replacing some of the category stalwarts.
"Rather than having them go do their call recordings, review call recordings, but then jump to other platforms to do really tangential things — we have them do it all in the platform where they're doing kind of the job to be done." — Brandon Bussey
The practical example: call prep agents built directly inside Glyphic/Airspeed that pull Salesforce data and prior call notes to generate a prep sheet before every call. It's a workflow that many teams are solving with a standalone Claude or ChatGPT agent. The difference isn't capability — it's adoption. An agent that lives inside the tool a rep already opens every day will get used. An agent that requires a separate prompt interface probably won't, especially for team members who aren't deeply AI fluent yet.
This is a point worth internalizing for anyone currently building AI tooling for go-to-market teams. The question isn't just "does this agent work?" It's "does this agent work where the rep already is?" Those are different questions with different answers.
There's a counterintuitive thread running through this conversation: sometimes the right move is to make it harder to build things, not easier.
Brandon describes a former Salesforce admin who would run every build request through a business impact checklist before touching anything. It was frustrating in the moment. It was also right.
"I don't know, seventy percent of the things I was gonna ask for — we didn't build because it's like, yeah, okay, I'm convinced that's actually not worth it." — Brandon Bussey
The same principle applies to agent development. The ease of vibe-coding and spinning up automations has removed friction that was doing useful work. When something takes a week to build, you think harder about whether to build it. When it takes an afternoon, you build first and question later — or never. The Go-to-Market AI Specialist role Brandon is hiring for is partly a technical function, but it's also an intentional friction mechanism: a human checkpoint that makes teams articulate the business case before committing resources to building something.
This connects to a broader question about what actually separates RevOps teams that drive outcomes from those that just drive output. The answer has never been purely technical capability. It's the discipline to say no to the things that don't move the needle, and to build accountability structures that hold up as the team scales.
The AI SDR project at Advantive is Brandon's clearest example of a lesson learned. The team brought in a third-party vendor, scoped an ambitious use case, and ran into the classic failure mode for AI projects: trying to solve too much with a single agent instead of breaking the problem into smaller, discrete tasks.
"Where it gets really good is when you have an agent given a very specific task. 'Okay, do this' — then another agent does this, then another agent does this, and then you can have an agent that then runs those agents." — Brandon Bussey
The AI SDR project is back to the drawing board, with a fresh build-versus-buy evaluation underway. The framework Brandon applies to that decision mirrors what Kyle Norton (CRO at Owner) described at a recent RevOps event: don't buy things that sit over your core intelligence layer. If data access is the strategic asset, you need to control it.
For Brandon, the call recorder decision is illustrative. Glyphic/Airspeed holds call data that drives enablement decisions, strategic direction, and go-to-market choices across the business. The requirement isn't just that the tool works well — it's that the data is accessible, queryable, and available for analysis across other systems. A vendor that treated call data as its competitive moat would be a non-starter.
The ROI calculus for the rest of the stack is simpler. Running lean teams means opportunity cost is always on the table. Rebuilding a call recorder from scratch when one already exists — even if technically feasible — is a poor use of the same engineering capacity that could be building higher-value differentiation. This is the kind of thinking that underpins effective RevOps prioritization at any scale, and it doesn't change just because AI makes building faster.
Ask Brandon to get out his crystal ball and he does it with a finance lens, which is telling. The framing isn't "AI will transform go-to-market" — it's "investors are going to demand that the productivity gains from AI show up in the numbers."
"The amount of output and quality output that just any knowledge worker can produce should be 10, 20X. And sure salaries and the expenses will adjust to an extent, but the reality is you should be able to drive much higher profitability as a result." — Brandon Bussey
The Rule of 40 is his reference point — the software industry's longstanding benchmark for balancing growth and profitability. His suggestion isn't that it becomes the Rule of 100 overnight. But the direction is clear: investors are already asking for more profitability alongside revenue growth, and AI-enabled productivity is how that math gets done. Teams that can't operate in that model will face pressure. Teams that can will have a structural advantage.
What separates winners from laggards in this environment, Brandon argues, isn't a particular tool or strategy — it's the willingness to revise your approach on short timelines. The companies that succeed will be the ones that look back every three to six months and say, honestly, "our thinking has changed." That's genuinely hard to do, especially in organizations with established playbooks and leadership teams that are used to longer planning horizons.
As Matthew puts it, the golden era of SaaS created a long run of relative stability where the same playbook worked across companies and years. That era created habits that don't serve operators well when the environment is changing in months, not years. The operational muscle that matters now is a willingness to be wrong about your current approach — and to update without ego.
For RevOps practitioners navigating the question of what their role looks like as AI takes on more of the execution layer, Brandon's take is grounding: the jobs become more critical, not less. AI multiplies what a person can produce. Humans still have to decide what's worth producing, who owns it, and whether it's still working six months from now.
Check out our blog, join our community and subscribe to our YouTube Channel for more insights.
Our average member has more than 5 years of RevOps experience. That means you’ll have real-time access to seasoned professionals. All we ask is that you’re generous with your knowledge in return.
