The assumption most people make
When people talk about AI in sales, they almost always default to high-velocity. Volume play. Automate the list. Generate the emails. Send more. Move fast.
And when you push back on that, someone will inevitably say: "Well, enterprise is different. It's too complex. Too relationship-driven. AI doesn't really apply there."
Both assumptions are wrong. And they're wrong for the same reason.
I've spent nearly a decade running a 70-person sales consultancy helping over 100 teams go to market — everything from transactional SMB deals to 18-month enterprise cycles with 12 stakeholders and a procurement team that wanted a referral from a competitor. I've seen both motions up close. And the teams that win in both of them share one thing in common: they're disciplined about what humans should do and what shouldn't require a human at all.
The teams that lose? They either automate everything and wonder why their domain reputation collapsed. Or they do everything manually and wonder why their best reps are burning out on admin instead of selling.
Let me take these one at a time.
What high-velocity actually requires
High-velocity isn't a volume game. It's a quality-at-volume game. There's a difference, and most teams miss it.
When I was at Trustpilot as the first sales hire, I closed nearly 600 deals in two years. Almost one deal a day. The company went from 50 to 5,000 customers in that time. People look at those numbers and assume it was pure hustle. It wasn't. It was research.
Before every single call, I would look at the prospect's checkout funnel, their social media presence, their online reputation, how they handled customer reviews. By the time I picked up the phone, I wasn't guessing. I already knew their world. The first 20 seconds of every call landed because I had something real to say. And something real to say is what turns a cold call into a peer-to-peer conversation.
Now multiply that by 80 accounts a week. You can't. Not manually. Nobody can. So what happens? Reps stop researching. They filter by employee count and industry — the only levers available in most databases — and they call the list. Every rep at every competitor is filtering by the same levers. They're all calling the same accounts with the same message.
Filtering by employee count and industry gets every rep to the same list. The research score is what separates yours from theirs.
The problem with high-velocity isn't the pace. It's the prioritization. At any given week, inside your TAM, there are accounts signaling readiness right now — a new hire, a strategic initiative, a leadership change, a funding announcement — and accounts that look identical on paper but aren't moving. Without research, you can't tell them apart. You call everyone the same way and you get the same mediocre results at scale.
What AI augmentation actually does for high-velocity teams is solve the prioritization problem. Agents score every account in your TAM against your ICP criteria, then research each one and return a research score based on how many of your defined signal questions came back positive. Two companies with identical firmographics can score completely differently based on what's actually happening inside them right now. The highest scores go to the top of the queue. Agents then find the right contacts using persona cards — not job titles — and generate outreach built from what was found, not from a template.
That loop runs every week. Automatically. Without anyone asking.
Reps start Monday with a ranked list. Not sorted by company size. Sorted by how much buying context agents found. The research I did manually for 600 deals at Trustpilot — they get it done for them, at scale, before the week starts.
That's augmentation. The judgment is still human. The question "is this the right company to call this week?" is now answered by context, not instinct.
Jonathan Wuurman, VP of Growth at Luzmo, inherited a CRM with 278 contacts when he joined. After running them through qualification and persona validation, only 4 were still relevant. Agents rebuilt the buying groups to 73 verified contacts. 6 SQLs in the first 4 weeks. "Accurate data is much more important than a lot of data which is not qualitative." — Jonathan Wuurman, VP of Growth, Luzmo
What enterprise actually requires
Enterprise sales teams hear "AI" and think it doesn't apply to them. The deals are too complex. The relationships are too important. You can't automate a 240-day cycle.
They're right about one thing: you can't automate a 240-day cycle. But that's not what augmentation means.
Here's what actually kills enterprise deals. Not bad reps. Not bad products. Research going stale between touchpoints.
A 240-day deal has 30 or more conversations. In between each one, the account changes. Leadership moves, strategic priorities shift, budget gets frozen and unfrozen, your champion gets promoted or leaves. The rep who knows what changed since the last call walks into the next one with credibility. The one who doesn't walks in with a pitch that no longer matches the room.
Tom Burg, SVP Marketing at Aqfer, told me he used to do account research himself before major calls. And by the time he finished one account, he didn't want to do another for months. The depth of research a complex deal requires is so demanding that most teams just don't do it consistently. They rely on what they remember from the last conversation.
Sven Roeleven, SVP Solution Management at ARIS, embedded Evergrowth into their global GTM enablement strategy. The goal was to find new buying signals and act on them with personalized, differentiated messaging. "The way to do outbound has changed dramatically. The combination of legal constraints and changing buyer behavior together made traditional outreach ineffective." Once the team uploaded their value proposition, ICPs, and personas, agents began finding the right contacts and surfacing buying signals for account planning. Account research that previously took hours now takes seconds. "Now with agentic AI, the platform does the work for you 24/7. That's the game changer." — Sven Roeleven, SVP Solution Management, ARIS
The other thing that kills enterprise deals is buying committee blind spots. You're rarely selling to one person. Economic buyer, technical evaluator, champion, legal, procurement — each with different concerns, different timelines, different objections. Reps running manual research can map two or three at best. The rest are people they don't know are in the room until they're suddenly blocking the deal.
And then there's the champion problem. Your champion was the person who understood your value, advocated internally, and had the relationships to push the deal through. When they leave — and they always leave eventually — the deal doesn't automatically die. But it will if nobody rebuilds the relationship at the new company and replaces the champion internally. Most teams find out a champion has left when the deal goes dark. By then it's too late.
What AI augmentation does for enterprise teams is handle the intelligence layer — the continuous monitoring, the scheduled research, the account planning — so that reps can focus on what only a human can do: the relationship, the strategy, the judgment calls that move a complex deal forward.
The rep still runs the deal. The agents make sure they're never walking into a conversation blind.
Why AI augmentation works for both
Two completely different motions. One thing in common.
Both require human judgment on top of research. The research layer is where agents belong. The judgment layer is where humans belong. The mistake most teams make is letting the research layer eat the time that should go to judgment.
In high-velocity, the rep's judgment is: which of these accounts do I call first? What angle do I take? How do I open the conversation given what I now know about this company? That's irreplaceable. What agents replace is the hours of list-sorting and manual research that currently happen before any of that judgment can be applied.
In enterprise, the rep's judgment is: how do I advance this deal? Who do I need to build a relationship with? What does this account's new strategic priority mean for my positioning? What do I do now that my champion has moved? That's irreplaceable. What agents replace is the 4 to 5 hours of manual research per account that currently has to happen between every significant touchpoint — research that, if it happens at all, often happens at 11pm the night before a big call.
Agents do the work that shouldn't require a human. Reps do the work that can only be done by one. The division of labor is the strategy.
There's one more thing both motions share: they require a trained agent, not a prompted one. Reaching for ChatGPT or Claude before a call to "research the account" is not the same thing. A general-purpose AI only knows what you paste into it. It has no ICP. No signal library. No awareness of what changed at this account since your last conversation. No persona cards. No playbook. It produces plausible-sounding context, not accurate context.
That distinction matters enormously in high-velocity, where you're making prioritization decisions across hundreds of accounts. And it matters even more in enterprise, where a single bad piece of intelligence can cost you a deal that took six months to build. The comparison is worth understanding.
Why AI automation kills both
Now for the part that frustrates me.
Every week I see someone on LinkedIn bragging about sending a million AI-generated emails and getting 78 meetings. That's a 0.0078% conversion rate. And the other 999,922 people? They now associate your brand with spam. Your domain reputation is degraded. Your reply rates will keep dropping. And the next campaign will need two million emails to get the same 78 meetings.
I'm not moralizing. I'm doing math. It doesn't scale. It doesn't work. And it actively destroys the channels that still do.
I've watched this happen to every outreach channel in B2B sales. Phone worked until everyone automated it with robocalls. Email worked until everyone automated it with sequences. LinkedIn worked until everyone automated it with connection request spam. The channel doesn't die because it stops working. It dies because automation floods it with noise until the signal disappears. We're watching it happen to AI-generated email right now, in real time.
For high-velocity teams, AI automation is particularly seductive because it looks like the obvious solution. Volume is the goal. AI generates volume. Connect the two, press go. But volume without prioritization isn't high-velocity sales. It's spam with a CRM attached. The accounts that would have converted get buried in the noise with the ones that never would have. And your domain, your brand, and your team's reputation take the hit.
The difference between data-driven and context-driven outreach isn't a nice-to-have. It's what determines whether your outreach gets read or gets filtered. Two companies with identical firmographics get the same automated email. Two companies with different research scores get completely different outreach built from what agents actually found. One approach generates noise. The other generates conversations. The difference between data-driven and context-driven outreach is the whole game.
For enterprise teams, the calculus is even simpler. You cannot automate a relationship. The moment a prospect realises they're in a sequence — and they always realise it — the deal is over. Champions don't advocate for vendors that treat them like a data point. Decision makers don't trust companies that clearly don't know their world. The relationship is the moat in enterprise sales. Automation removes the moat.
I've seen enterprise teams try to "scale" their outreach by automating touches in between calls. Generic check-ins, automated follow-ups, AI-generated "just circling back" messages. The effect is always the same. The prospect starts responding less. The champion gets harder to reach. The deal quietly dies while the rep is confused about what happened. What happened is that the relationship was replaced by a workflow. And no one buys from a workflow.
The right division of labor
So here's how I think about it.
In high-velocity: agents score your TAM, research every qualifying account, find the right contacts using persona cards, and generate outreach from what they found. Reps review the queue, make the calls, have the conversations. The loop runs every week without anyone asking. Reps start Monday ready to sell, not ready to research.
In enterprise: agents monitor every active account continuously, resurface fresh intelligence before every touchpoint, build the account plan from the research, and map the buying committee across the full stakeholder landscape. Reps walk into every conversation briefed. They never get blindsided by something they should have caught. They focus entirely on the relationship and the deal strategy.
In both cases: agents do the work that shouldn't require a human. Reps do the work that can only be done by one.
The teams that figure out that division of labor will outperform the ones that don't. Not because their AI is smarter. Because they stopped asking their best people to do work that doesn't require their best people. And they stopped asking AI to do work that requires a human being.
That's not a technology strategy. That's a management decision.
If your AI strategy is to generate more volume on channels that are already overcrowded, you're not augmenting your team. You're accelerating the destruction of the channels that still work.
The tool doesn't matter. The motion doesn't matter. What matters is whether you understand which work belongs to humans and which work shouldn't require them. Get that right, and the rest follows.
Frequently asked questions
Does AI in sales only apply to high-velocity teams?
No. Both high-velocity and enterprise teams benefit from AI augmentation, just differently. In high-velocity, agents solve the prioritization problem — scoring TAM, researching accounts, and generating outreach so reps start the week with a ranked queue. In enterprise, agents handle the continuous intelligence layer — monitoring accounts, refreshing research between touchpoints, and mapping the buying committee — so reps walk into every conversation briefed.
What is the difference between AI augmentation and AI automation in sales?
Augmentation means agents do the work that shouldn't require a human — research, prioritization, account planning, contact mapping — so reps can focus on judgment, relationships, and conversations. Automation means handing the entire outreach process to AI, which floods channels with low-quality volume, destroys domain reputation, and breaks the trust required in enterprise relationships.
Can you automate enterprise sales with AI?
You cannot automate a 240-day, multi-stakeholder cycle. The moment a prospect realises they're inside a sequence, the deal is over. What AI should do in enterprise is keep research fresh between touchpoints, surface buying committee changes, and maintain account plans — so reps can focus entirely on the relationship and the deal strategy.
Why is using ChatGPT or Claude for account research not enough?
A general-purpose AI only knows what you paste into it. It has no ICP, no signal library, no persona cards, and no awareness of what changed at this account since your last conversation. It produces plausible-sounding context, not accurate context — fine for inspiration, dangerous as a basis for prioritization or for a complex deal.
