The AI Slop Problem: Why More AI Is Making Sales Teams Worse

Every revenue team we meet has added AI to its stack in the last two years. Call summaries arrive seconds after a meeting ends. Follow-up emails draft themselves. Dashboards light up with alerts. And yet, in the deal reviews we sit in, the most common complaint from sales leaders is not that they lack information. It is that they are drowning in it.

We call this AI slop: a high volume of plausible, generic, single-conversation output that looks like insight but rarely changes what anyone does next.

Twenty things became sixty

One sales leader described it to us simply. Before AI, there were twenty things on their plate. Now there are sixty. Every call produces a summary, every summary produces action items, and every action item competes for the same finite hours of the same people.

The problem isn’t that the tools are wrong. Most of what they produce is technically accurate. The problem is that accurate and useful are not the same thing. A summary that faithfully records what was said in a 45-minute call tells you nothing about whether the person who said it can sign the contract, whether their boss has ever heard your name, or whether the objection they raised was already raised, and ignored, three calls ago.

Accurate and useful are not the same thing. Slop is output that is technically correct and strategically irrelevant.

The race to the median

There is a second, quieter cost. When every sales organisation prompts the same frontier models with the same transcripts, every team gets the same kind of answer: the same tidy recap, the same polite follow-up email, the same generic list of “next steps”.

That output is, by construction, average. It reflects the statistical centre of how deals are usually discussed, not the specific, often counter-intuitive moves your best sellers make. Top performers notice when a technical stakeholder starts asking circular questions. They know when to stop demoing and start whiteboarding. They sense that a “positive” call was really a polite exit.

Generic AI flattens exactly that nuance. Mid-tier sellers get a little better; top performers get pulled toward the middle. Across a team, the result is a race to the median, and in competitive enterprise markets the median does not win.

Single calls never tell the story

The deepest limitation is structural. Most AI tools in the revenue stack evaluate one conversation at a time. But a seven-figure deal is not a conversation. It is a months-long, multi-threaded negotiation across a buying committee of a dozen or more people, spread over calls, email threads, support tickets and side conversations you never see.

The truth of a deal lives in the connections:

• the champion who is enthusiastic on every call but has never mentioned budget;
• the economic buyer who has not appeared in three weeks;
• the security objection that surfaced in an email, resurfaced on a call and is quietly hardening;
• the procurement timeline that only makes sense if someone else is also being evaluated.

None of these show up in a summary of any single meeting. They only appear when you read the whole account, over time, and attribute each signal to who said it and when.

What good looks like

The answer is not less AI. It is AI pointed at the right problem, paired with judgement, and finished all the way to action. In practice that means four shifts:

1. From volume to signal. Fewer insights, ranked by how much they change the outcome of the deal, not a list of everything that was said.
2. From single calls to the account. Reasoning across every conversation and channel in a deal, so patterns that only emerge over time become visible.
3. From assertion to evidence. Every finding traced to the buyer’s own words, with a citation a manager can check in seconds. If you can’t audit it, you can’t forecast on it.
4. From insight to execution. The output is not a dashboard. It is the email to the CFO, the whiteboard agenda for the technical team, the one-page answer to the security objection, ready for the rep who owns the deal.

That last step is the one most organisations skip, and it is where most of the value sits. An insight that dies in a dashboard is, at best, another item on the plate.

A five-question test for your stack

If you are not sure whether your AI investment is producing signal or slop, ask your team these questions about your three most important open deals:

1. Who has final authority over the budget, and when did they last engage?
2. Which objections have been raised more than once, and has anyone closed them in writing?
3. Is our champion carrying our message internally, or only agreeing with it in front of us?
4. What does the buyer believe it costs them to do nothing?
5. What is the single most important thing we should do this week, and is the asset to do it ready?

If your tools can’t answer these, with evidence, in minutes, they are adding to the noise.