Inside Jeff Ignacio’s RevOps AF keynote: The future of RevOps with AI

Speaker A: M. Uh, uh, um. Welcome to the RevOps review, the podcast unpacking what really works in modern revenue operations. I'm Jeff Ignacio and this is a podcast from Cognizant, helping revenue teams run on accurate, compliant AI ready B2B data they can actually trust. Let's get into it. Hey everyone, it's Jeff Ignacio, the host of the RevOps Review. And today is going to be a solo episode. Normally I bring on guests so you get to hear their perspective. And I love sharing the perspective of my guests. However, this week I want to focus on something different. There was the RevOps AF conference, the annual festival. That's what AF stands for. That doesn't stand for anything else. Despite what you might think by the RevOps co op community, I was honored and privileged to be the closing keynote for the event. And so what I'd like to do today is actually walk you through my presentation and for those who can't see the screen on my screen, share. That's okay, I'll walk you through it. And so everyone knows this, but I have been a big believer that AI could replace much of what we do in review operations. But quite frankly, after experimenting with AI for quite some time now, I've come to change my view and I'd like to share a little bit about that perspective. So the presentation that I had this week was called the future of RevOps with AI. And so one of the things that most folks like to think about is this one question of, uh, will AI take my job? And I personally think it's the wrong question because when we think about how we think, you can either comment things in life with a defensive question or be more proactive in life. And so a defensive question, they'll actually produce what I call defensive thinking. And when you have defensive thinking, thinking, it's actually what causes us as revenue operations practitioners to miss the most significant probably change, not incremental, but an exponential change in our function probably for the next generation, the next 20 years. And RevOps has changed so much in the last 10 years. I mean when I came into the profession, there were new tools around sales engagement, how you can sequence relevance versus personalization became the conversation point for go to market tools. We had tools that were vertically integrated for business intelligence solutions. To see your funnel from end to end, connecting to the top of funnel all the way to sales execution to the post sales side of the house. And that's what we do in revenue operations. This is continually craft at tying the funnel together and generating alignment across our stakeholders. Here comes AI on both the automation side of the house, the LLM and agentic capabilities, it's exploding the possibilities of what AI could provide. And so there's one theme that I think is interesting. When we think about doing more with less is somewhat the prevailing theme that we've heard about in 2026. There's uh, another thing, another way to think about it and it's called Jevons paradox. So when we think about Jevons paradox, it's a theme that came around a very long time ago. 1865, William Stanley Jevons observed that as, uh, steam engines became more efficient, coal consumption didn't actually fall. In fact, consumption exploded. And why is that? Because more efficient engines actually made coal economically viable for uses that didn't previously exist. So imagine how that applies to revenue operations. AI pushes the frontiers of what we could provide and instead of doing less, we're actually asked to do more. And I think that's really interesting. So more efficiently, more efficiency doesn't reduce demand, it actually creates demand that didn't previously exist. And so when we think about, you know, teams using AI to do what they already did, just faster, those are really incremental gains. But I think what we want, what I would like to focus on is a, ah, radical repositioning of revenue operations. It's where teams using AI can do what was previously impossible. And that, that's a completely different revenue operations function. And so I'd like to continue focusing on this second group. So you ask yourself what becomes possible. And in my mind, in some of the implementations that I have done personally, you know, I'm going to walk through five capabilities that I thought were theoretically, you know, theoretically possible but impossible, implausible to actually execute, but now becomes available because of AI? The main constraint was bandwidth, really headcount and budget. So the constraint was always the same, it was human bandwidth. So here's one example. Example number one. Example number one is full pipeline coverage. A real life example of this is that a Manager reviews maybe 20% of an active pipeline in any given week. And in if you have a wide span of control, a sales manager has eight reps and the eight reps probably have five to 10 meetings a week. So you're talking about five to 10 times eight. So 80, 40 to 80 calls a week, let's say 50% of those convert into a uh, qualified opportunity. So now you're talking about 20 to 40. Is it realistic that a manager is going to review 40 net new opportunities on their team in addition to the in flight Opportunities? I don't think so. So, you know, the deals that get attention are, uh, quite frankly the loud ones. And when I say the loud ones, they're the ones that commit. They're the ones with the close date that's coming soon. But now with AI, the other 80% that were previously invisible, they can actually get surfaced. They get the light, the day and the light that they deserve. And so every deal is reviewed against the same criteria every week. So things like stage duration, activity cadence, engagement signals from the buyer side, you know, you're not looking at a sample of data, you're actually looking at the entire population of your pipeline. And so that's what AI could do. And it could do it at scale. Now example number two is one that I love. This is a win loss analysis. So imagine, well actually you don't even have to imagine. I bet you're running your win loss analysis every quarter. And if you're not doing it, you're maybe doing it once a year as some sort of product marketing, uh, project internally. But when you have a closed loss reason field, you know, we all are coding, you know why we lost, and it's a drop down and you might have an open text field, but by the time anyone sees the output, those signals are 90 days old. In fact, the win loss deck is, you know, is in a folder that no one probably reads. And so with AI, we change that. Every call transcript, every pricing objection, every competitive mention, it runs through a continuous analysis. And now you can see that patterns will surface in days and the positioning changes, you know, the updates that you have to your positioning, they actually happen in weeks. So it's this feedback loop from the field to enablement to product marketing, back to the sales floor to execute. And that feedback loop compresses over time. It compresses from quarters to weeks and uh, maybe even faster than that. So real time coaching. Now the third one is scenario modeling. So when we think about our plan, the annual plan is set in stone at the end of the fiscal year and put into motion and commemorated with the kickoff at the beginning of the fiscal year. Many of you are probably wrapping your Q1, uh, probably when you listen to this episode and the thought is, are we ahead or are we behind plan in terms of pacing quarter to date relative to where we should be for the year? And then do we have sufficient pipeline for not only the next quarter, but do we also have the campaign plans and the headcount hiring plans to make sure that we generate sufficient pipeline for the back half of the year. And so when we think about scenario modeling, you might do things like a uh, territory redesign, new headcount plan, you might have new product releases and so scenario modeling can actually happen in hours. Now an LLM isn't so great at modeling, that's for sure, but you could use maybe a coding agent of some sort to develop discrete models that are grounded in map with appropriate ranges and those ranges not also paired with different scenarios. You might have a base case scenario, a stretch scenario, a conservative scenario. Those scenarios might include differences in average pricing, difference in win rate, difference in sales cycle. You might also throw in you know, new products that are released, new regions that are released, headcount ramp. So all those factors come in and you can start to really put together. You know, in my world I like to think about a uh, Monte Carlo analysis. What if you can run the quarter 10,000 times and put those ranges for each variable on uh, the high and the low side and you can really run a scenario around how would the year play out? And you can get that, you can get to that, those data points much faster to surface to build your models. I think that's one of the things that when we think about, you know, native AI built for the future revenue operations, that that's an interesting use case. Now two more use cases. Use case number four is rep coaching at scale. Many of you may have a conversational intelligence solution. Listening to what happens in every single sales call is very timely. It's time consuming. A manager with eight direct reports can probably review maybe two to three calls per rep per week. Same with the pipeline example for example number one. And so the coaching is going to be biased to the reps who are struggling or maybe even the loud ones. So the middle of the distribution gets pretty much no support. And so every call reviewed with AI, the talk ratio your question density objection, handling your next step commitment. For every rep every week the manager's job is going to shift from observation to intervention taking action. So the middle of the rep distribution starts to become visible. And quite frankly when we think about solving for sales, we're always trying to move the average or the mean of the middle band of our rep performance. So we're now able to focus not on the wings, not on the tails, right? The excellent performers and the poor performers. But we're actually spending our time in the middle. And then lastly, churn prediction. This is going to require some tunneling, piping data from product usage metrics, getting that those signals into your CRM. So yes, AI is not going to Be as helpful until the data plumbing is in place. But you know, churn detection, you know, historically has been reactive usage drops. Your CSM might notice this a couple of weeks later, maybe the time, by the time that flag goes up and it sent to the team, the customer's already decided you know what they're going to do internally. And at that point it becomes a save play and your percentage of actually turning it around is going to be minimal. And so afterwards, once you build your data pipelines, you can then surface those data points as inputs to your AI model, which would then allow you to look at product telemetry, support volume, NPS scores, exec engagement, renewal, timing. All this combined into a continuous risk model. Not just a one time risk model, but it's continuous. And then your intervention can happen while there's still sufficient Runway to change the outcome. So a couple of things there I think are super interesting for use cases. So when we think about applying Jevons paradox, you know, AI actually doesn't shrink revenue operations. It expands what revops is accountable for and it permanently changes the frontier, like what skills we need in order to operate. Now, when a function becomes more capable, the organization demands more of what it can now do. So scenario modeling on demand, deal intelligence at pipeline scale, continuous win loss analysis, all. Every time you deliver more value, I always find that my stakeholders always want more. When you deliver the goods, they're going to want more of it. But when you deliver more goods, again, they're going to want more of it. So we start moving from reactive to maybe proactive. That would be the ideal. And then beyond. Proactive is also agentic. Now agentic is interesting because you can't compete with it as a human. The agentic layer runs 24,7, but it has to be programmed and thought through and scaffolded by a human. I'll walk through why. So the team doesn't shrink. But I will say the composition of what we do in revenue operations will fundamentally change. So before AI and after AI, I'll position this next set of thoughts. On the data side, before AI, analyst who pulls, reports and builds dashboards manually. After AI, we architect, who designs the signal layer that the AI will query on the ops side of the house. Before AI, the admin will configure the rules, the sequences and routing logic. But after AI, we become operators who instruct, monitor and audit the agent. And then on the scoping side, before AI, the generalists who covers everything at lower depth. Then after AI, we are now the specialist who owns evaluation and quality control. So the ceiling on what a small rev ops team can be accountable for has fundamentally moved and has moved to the right. So what skills will win tomorrow? So if you're just stepping into revenue operations, you're going to walk into a world, quite frankly, that's going to be a fun learning journey for you. For those of us who've been in the space for 12 to 15 years like I have, it's quite frankly going to look different. So what new skills do we need to build and then what skills do we already have that are timeless and compound? So the combination of these two is what the practitioner who's genuinely hard to replace. So when I think through important skills, prompt and workflow design is the first skill. And if we've been prompting since 2025, clever phrasing like my job is on the line, we're shifting that to strategic structuring. Understanding the reasoning process of a prompt. You may have single prompted before, but now you're moving to a series of compound prompts or explicit frameworks such as a, uh, give it the context, give it a task, tell its constraints and then give it an output you may have had. Maybe conversational prompt engineering. You're asking one question at a time, building up slowly over the course of a session with an LLM. But now we're moving to a cognitive architecture. You're really thinking through the workflow of what an AI agent should build towards. Also the AI agent in 2025 we would ask a question and what would come out would be magic. It'd be long winded text. Natural language outputs. Natural language outputs are hard to work with. So if you're working with workflows, revenue operations, you want to move to an API contract format where it's a JSON output. It's a little technical, but at least it's going to be queryable and actionable and consistent over time when you're working with your AI operations team. Second thing is thinking through agent architecture. Now you don't need to build it yourself, but in revenue operations I highly suggest you need to evaluate it, spec it, you know, build the specifications and know when it's broken. So when you're thinking through the multi agent architecture, you still need to know what good looks like. How does a VP of RebOps stay relevant in the age of AI? Well, guess what? Your experience is all super relevant because you know what good looks like, right? When you see an output that AI produces and it's off the mark, you need to know why you need to Dig into the architecture and go through each step of the agentic workflow and see where the thinking was broken. Change it so that the outputs match exactly what you're looking for next is the evaluation design. How do you know your AI output is correct? In fact, again, I'll reiterate this. The most underrated skill in the room is that you want to design spot check protocols. You want to create confidence thresholds and a human in the loop review so that it's a gate that sits between the AI output and the executive decisions. You do not want to just print an AI output and send it to the boardroom or to the executive team. You need to be able to inspect what's happening. Lastly, I think signal architecture is important when we think about inbound and outbound. You need to be able to capture signals and time to value is so important here. Time, speed to market. So what data do you feed the model and why? The practitioner who understands which signals actually predict the closing deal and can translate that into a data model that the agent runs against is just not replaceable. They are not the reason that AI produces good inputs. So what does good look like? That's when you select when you have selecting selective signals for a churn model product, telemetry, support, volume, exec engagement, NPS renewal timing. Each of these signals has a weight and a cadence decision around it. And that decision then requires GTM knowledge, not just data. You need to be able to know what to do with these signals. So the only way to check an AI system is to know what good looks like. Independently. Many of the skills that we as revenue operators have cannot be replicated. In fact, I would say this humans don't necessarily need AI, but I would say that the AI work needs humans to look at to review it. So you want to make sure that the outputs you're getting from AI are trustworthy. And that's what makes the powerful. That's what makes for a potent combination between this human and this AI interface as we move into this new world. So let's go over some timeless skills. I'm going to leave this with you. So there's four timeless skills that I think are going to be extremely relevant. First is business acumen. So the four pillars of business acumen here are going to be 1 financial literacy, 2 market analysis, 3 operational knowledge and 4, strategic thinking. So an AI pipeline summary can tell you what the numbers are. But when you apply those four pillars of business acumen, it you know what they mean. Is it seasonal, is it structural, is it A rep problem or a positioning problem, what's noise and what's signal. The model surfaces the pattern, but the model may not interpret the pattern correctly. And so when the human reviews the output, you're able to understand what the pattern implies and then narrate back to the business. Second is go to market acumen. So you know, going a little deeper into our own functions. Marketing, sales, customer success, pre sales. If you've ever built a capacity model from first principles, you cannot just, you know, catch the AI's assumptions error right away. You have to be able to see, for example, if you had AI build a territory model, you have had to maybe build a territory model yourself by hand. You know, what makes sense to you, what does territory covers look like, how many accounts should they work, uh, what's the mix of Tier 1 versus Tier 2 versus Tier 3? So when we think about it, your go to market knowledge actually becomes the review layer, the review layer again of uh, AI outputs. Third, for those who are technical savants in HubSpot or Salesforce or any of these other tools, your fluency in these systems actually superpower the outputs from AI. So if you're using for example a CLAUDE and you're just chatting into it, claude's going to default to the mcp. It's uh, a way for agents to interface and interact and pull data from your systems. Now you may not know this, but the MCP out of the box is only going to pull, you know, 200 records or 2,000 records. It's not going to pull the full sample. When you're running an analysis on your own, what do you do? You make sure that you're filtering correctly and getting the full data set. And you would only know this if you had done this before by hand. And so your ability to work with data models, systems and then surface that through AI agents is going to be powerful. So in this world of AI and rev ops, your skills actually don't decay, they actually become even more prominent. Lastly, I would say stakeholder communication. AI is not going to run your qbr. It's not going to present to the board, it doesn't navigate the organizational dynamics of a territory redesign. You know, this account moves from west to east because the person who's been supporting that account is moving regions. It's not going to know that the judgment about what to surface, how to frame it, who needs to be in the room and what trade off to recommend. I personally think that that remains entirely human. And so the human, the human here builds trust, particularly with table stakes decisions. So the combination that wins is going to be combining those new skills and the timeless skills. And when you do that, I personally think you're going to be the practitioner who is genuinely hard to replace at your org. So just last end of this, again, solo episode on AI and revenue operations. The future of Revops is that Revops is not going to be smaller. In fact, I think, I think it's going to be even more, it's going to be even larger than before, going to be more strategic, more instrumented, more accountable and more consequential than it has ever been before. So I'll leave you with this. You know, go build at that level. Good luck. Go forth and operate and good luck out there. That's it for this episode of the Revops review. Subscribe for more and if you're enjoying the show, a quick, positive review really helps. Thanks for listening.