A salesperson on £125K basic, probably clearing a quarter of a million with commission, is sat there scraping LinkedIn and building lists in a spreadsheet.
Every founder and CRO who sees this has the same instinct: automate it. That’s the most expensive hour in the building being spent on the cheapest work. Rip it out. Buy the tool. Give them their time back.
I get the maths. I’ve done the maths myself as a CRO three times over. But the instinct is wrong, or at least half wrong, and it’s the wrong half that does the real damage over time.
The research is what makes the rep good
Here’s what I’ve seen consistently, both when I ran revenue teams and now working with founders and CROs on their GTM motion: the reps who outperform do the hard work themselves.
They build their own lists rather than accepting what RevOps hands them. They go the extra mile on research rather than skimming the top few Google results. And when they get on a call, they’ve formed a genuine view of why this account, why now, and how they can win it – because they had to work it out themselves.
The tenured BDR beats the new one for exactly this reason. The tenured one cherry-picks. The new one gets fed a list. One is exercising judgement about who is actually winnable; the other is doing volume against accounts someone else told them to hit. The results follow the judgement, not the volume.
Same reason a much higher proportion of wins come from AE-sourced pipeline than BDR-sourced. The AE who built the account plan themselves has a much more developed sense of who the right accounts are, why they’re the right accounts, and what a real path to close looks like.
Strip that work away and you save time, but you kill the mechanism by which the rep gets better.
It’s like the famous ‘wax on, wax off’ montage from Karate Kid – doing the thing (over and over) – is good training that underpins great sales reps over the long run.
The ‘7 minute abs’ problem
This is the bit vendors don’t want to talk about. There’s a version of AI in GTM right now that is basically ‘7 minute abs’. Miracle drug. Do less, get more. Your reps can spend all their time selling because AI handles the research or ‘manual work’.
It’s like making the weights lighter or the runs shorter and flatter when working out. Superficially feels great. Easier day, more activity, dashboards look busy. Then you check in three months later and the team is in worse shape than when you started.
The pipeline is lower quality, conversion rates dropping. The reps have opinions about their accounts that are one layer deep. Discovery calls sound generic because the rep never had to think hard enough about the account to know what to ask. Losses get explained with the same three reasons every quarter because nobody is doing the pattern recognition work that produces sharper explanations.
The cost doesn’t show up in the AI bill. It shows up in win rates twelve months later and in a bench of reps who can’t tell you why a deal is real.
I see it in my own work too. When I just accept AI’s answer or the approach it suggests, the outcome is almost always worse than if I’d done the thinking myself and used AI as a copilot to sense-check me. The other way around – AI leads, I rubber-stamp – produces confident-sounding mediocrity. Every time.
The question worth asking
So the real question isn’t “how do we get the £250K salesperson out of the spreadsheet.” The question is:
Which parts of that process build intuition, and which parts are pure friction with no return?
Those are different problems with different answers. And conflating them is how you spend serious money on AI that makes your best reps worse.
Lets be specific about where the line sits, because vague framing is how vendors sell you the miracle drug.
Automate this because it doesn’t really sharpen thining:
- Manual gathering of information that’s already public. If your rep is copy-pasting from a company website into a doc, that’s pure friction.
- Contact enrichment. Finding the right people, getting emails and mobiles, mapping the org chart. Nobody’s judgement improves by doing this manually.
- Checking for signal alerts. New hires, funding, product launches, job postings, tech stack changes. The rep needs to know these have happened. They don’t need to be the one hitting refresh on Google News.
- Deciding which accounts matter and why. This is the account plan. This is the ICP judgement. This is the “why now” thesis. If the rep isn’t doing this work, they don’t have a real view, they just have a list.
- Interpreting what the research actually means. Enrichment tells you the head of ops changed jobs last month. Judgement tells you what that means for your deal and what to do about it. Different skill entirely.
- Forming a point of view on how you win. What’s the wedge. Who are the internal champions. What’s the likely objection sequence. What does the competitive landscape look like inside this specific account.
- Synthesising signals into a coherent story about the account. Raw signals aren’t insight. The synthesis is the insight, and the synthesis is where the rep’s brain has to be in the room.
None of this is AI, by the way. Most of it is closer to plain automation with a bit of AI at the edges. You’ve probably been able to do most of it for five years.
Don’t automate this, it’s where the rep actually gets sharper:
If the AE doesn’t know what they should be learning about their accounts, the research is just throwing shit at the wall. And if they don’t know how to interpret it, they’re outsourcing the critical thinking needed to win the deal and to become a better salesperson.
Where AI is most valuable
Once you’ve drawn that line, there’s a genuinely interesting role for AI on the thinking side of it. And it’s the opposite of what most vendors are selling.
AI as a thought partner. AI as the thing that challenges the rep’s assumptions, finds the holes in their account plan, stress-tests their “why now” thesis, plays devil’s advocate on their discovery hypothesis.
Not removing the thinking. Actually adding friction to the right parts of it.
The value isn’t “here’s your account plan, generated in 30 seconds.” The value is “you built an account plan, here are five reasons your thesis might be wrong, here are the questions you haven’t asked, here’s the competing narrative you should be pressure-testing against.”
That version of AI makes reps less lazy, not more. It makes them defend their thinking. It surfaces the assumptions they didn’t realise they were making. It’s closer to a good coach than a good assistant.
And weirdly, most of the AI budget in GTM right now is being spent on the wrong side of this equation. Big investment in tools that remove thinking. Almost nothing on tools that sharpen it.
What this means if you’re the one signing the cheque
If you’re a founder or CRO evaluating AI in your commercial function, the diagnostic question I’d run is this:
For each tool or workflow you’re considering, ask whether it removes the admin around a thinking task, or whether it removes the thinking task itself. Say the answer out loud. Be honest about it.
If it’s the first: good. Buy it, deploy it, get the time back, redirect the hours into the parts of the job that actually compound.
If it’s the second: understand what you’re doing. You’re trading short-term efficiency for long-term capability. Sometimes that’s a fine trade – a low-priority segment, a nurture motion, a market you don’t care about long-term. Often it’s not, and you’re eroding the very thing that makes your senior reps worth what you’re paying them.
The £250K salesperson doing list research sounds like a fail. But the failure isn’t “we haven’t automated the research.” The failure is that nobody has drawn a clear line between the parts of their job that build intuition and the parts that are pure drag. Until you’ve drawn that line, every AI purchase is a coin flip on whether you’re helping or actively harming the team.
Draw the line first. Then implement AI in the right spots.