Five AI Myths That Are Holding Your Revenue Team Back

At a recent RevOps Alliance Summit in Austin, Von CEO Sahil Aggarwal got on stage and told the room he was about to be "deeply unpopular." He then spent the next 20 minutes challenging five things that most people in revenue operations hold as truths. Some of what he said was uncomfortable for vendors. Some of it was uncomfortable for buyers. All of it was thought-provoking.

I've been working alongside Sahil for months as we bring Von to market, and these aren't new ideas for him. He says them in customer conversations, in internal meetings, at dinners we host. Hearing him lay it all out in one presentation, and observing the audience reaction, felt like the kind of talk people will reference six months from now and say, "Yeah, he was right."

Here's what he said.

Myth 1: "Our data isn't clean enough for AI."

This is the one Sahil hears the most. Every time we talk to a prospect about Von, someone says some version of “our CRM is a mess, our data hygiene isn’t there yet.” They don’t want to add AI and then deal with “garbage in, garbage out.”

Sahil challenged the room to recognize that it's just not true anymore. It's outdated thinking.

As an industry, we've taken decades to reach 80% data hygiene at best. We might never get to 100%. And the way AI is moving, we can't afford to spend two or three more years trying to get there. The AI models have gotten smart enough that you give them access to the raw data, and they figure it out. If an email came in at 2:47 pm and a meeting happened at 3:00 pm and a note was entered at 3:35 pm, the model will look at the context of each, match them to the right account and opportunity, and stitch the story together. It doesn't need you to have tagged everything correctly in the CRM first.

The data you need is already there. Your emails are in Outlook or Gmail. Your calls are (hopefully) being recorded. (If they're not, your sellers can take rough notes after meetings.) You don't need structured, field-level perfection. You need your unstructured data to exist somewhere accessible to AI. The right AI platform will do the rest.

As Sahil put it, “your CRO doesn't want to read through Salesforce records. They want to know which deals to call out.” With AI used the right way, it doesn't require a clean CRM and good data hygiene to produce this insight on-the-spot.

Myth 2: "If we just tell sellers what to do, they'll do it."

There's a persistent belief that if we build the right dashboard, surface the right next-best-actions, and serve up the right recommendations, sellers will follow them.

They won't.

We've all seen recommendations work on YouTube, TikTok, and Instagram. Those platforms have spent billions of dollars and have an almost incomprehensible amount of data on each individual user. The recommendations are truly, deeply personalized. But in revenue, we take a generic algorithm we think will work and give the same recommendation to every seller on the team. It fails.

Consider two enterprise sellers. One has won their last three deals because a champion switched jobs and brought them in. The other has won deals because they spotted spikes in website traffic and timed their outreach perfectly. These two sellers have completely different worldviews about what makes a good deal. Give them the same recommendation, and at least one of them will reject it, because it violates how they see their world.

What works instead is to build a base model of what good looks like, then let sellers layer in their own preferences. One seller might weigh job changes heavily. Another might care more about website activity or product usage signals. That middle ground between top-down recommendations and individual preferences is where adoption actually happens.

Myth 3: "We need to hire more AEs to increase revenue."

When a CRO needs to add $10 million in new ARR next year, the reflexive move is to hire ten more AEs. There's an assumption that one AE equals $1M in revenue.

Sahil pointed to Perplexity. They have hit a nine-figure revenue run rate with five sellers. The math from the old model won't be accurate for AI-forward teams. They can handle more.

This connects directly to an argument we wrote about previously: the right AI platform will dramatically expand what a good seller can handle.  A seller can only hold so many deals, so many relationships, so many competitive details in their head at once. When AI holds that context for them, the capacity equation changes completely.

Junior sellers and underperformers will have a harder time in this world. When quotas go from $1 million to $5 million or $10 million per rep, the bar for what "good" looks like goes way up. But top sellers with the right AI support will be able to handle dramatically more.

Myth 4: "We need an agent for everything."

The current market narrative is that you need AI tools and agents for every function. For example, one to update the CR, one to build decks, one for account planning, one for content creation, one for CRM administration, etc.

This is no longer true. The AI strategy that wins is when you have one agent that can serve your entire revenue team.

Think about Cursor (OpenAI's coding agent) and Claude Code (Anthropic's coding agent). They are not a collection of specialized agents. They are single agents that deploy their own sub-agents as needed, but for the user, it's a one interface with context, and can do any engineering task.

When you buy separate AI tools or create an army of separate agents for every function and for separate tasks, you create chaos by building in misalignment, repetitiveness with different answers to the same question, and often a graveyard of agents that aren't working as planned or expected. And that's not intelligence. That's confusion.

You may be thinking, "What does a graveyard of agents mean?" Sahil predicts that separate agents and AI tools for each function will play out the same way dashboards have. Companies will build agents. Those agents will break. Someone will need to fix them. If that person has left the company, the agent is dead. And instead of maintaining them, people will ping the ops team for help, exactly like they do today with dashboards.

What revenue teams need is a single intelligence layer across all go-to-market data. One agent who understands the business, remembers past interactions, and can handle anything from building a pipeline report to creating Salesforce flows to calculating commissions.

Myth 5: "We have too much AI."

Someone recently said to Sahil, "I already have too much AI. I don't need more AI." His response was to say, "AI is like electricity. Look around your living room. You have a TV, a Roomba, a digital photo frame, a speaker, a refrigerator, a washing machine, an oven. Every single one runs on electricity. Do you ever walk into your home and say, "I have too much electricity"? No. Because electricity isn't the problem. AI isn't the problem either."

You might have too many appliances. One is a refrigerator, an essential, you can't live without it. One might be a Roomba, nice but still not quite a must-have. And one is a digital photo frame. If it breaks, you won't notice.

The same is true for AI tools. The problem isn't that there's too much AI. The problem is that some of the tools you're buying are digital photo frames. The job is to identify which AI applications are essential to the quality of life and invest in them.

Why this matters for RevOps

One thing Sahil said that didn't get a myth number but deserves attention: every RevOps person he's talked to has been excited about what Von can do, because it means they can stop drowning in ad hoc requests that eat 50% of their day.

When sales leaders and managers can self-serve on dashboards, reports, account plans, and pipeline analysis, RevOps can get to work on more strategic projects like, “why do we stall  at stage two?” Or "What operational change would move conversion rates?" That kind of thinking used to take weeks because the execution ate all the time. Now it takes a few minutes or a few hours at most.

That's the future Sahil is describing. Not more tools, not more agents, not more dashboards. A single, intelligent layer that knows your business and lets every revenue function self-serve so the people who should be doing strategic work can actually do it.

We're currently offering free trials with enough credits to run through these use cases and pilot with your team. If any of this resonated, give Von a try.

Start your free trial at vonlabs.ai

Sara Kinsey is VP of Marketing at Von. Von is an AI data scientist for revenue teams, built by the team behind Rattle.