Why I Fired AI From Our Sales Process

Dennis Kriel • June 13, 2026

Share this article

About eighteen months ago I made a decision that confused most people who heard it. We had spent three months building out an AI-driven sequence for our sales outreach. Personalised messages, automated follow-ups, sentiment analysis on replies. The system was genuinely impressive. And we shut it down.

Not because it did not work. Because it worked too well in the wrong direction.

What we thought we were solving

The problem we were trying to fix was real. Sales conversations were taking too long to initiate. Our team was spending hours on research and personalisation that could be automated. We had good leads sitting untouched because there was not enough time in the day to reach all of them.

AI seemed like the obvious answer. And in theory, it was. The tools available now can do in seconds what used to take a junior salesperson an afternoon. We built the system, we tested it, and the output was slick.

But something was off.

The numbers looked fine. The relationships did not.

In the first eight weeks, our outreach volume went up by 300%. Response rates held. Booked calls increased. On paper, it was working.

Then I started sitting in on more of those discovery calls. And I noticed something the spreadsheet was not capturing. The conversations felt flat. Prospects had received a message that referenced the right things, used the right tone, arrived at the right time. But when the call happened, there was no warmth in the relationship. No memory of how we had got there. No connection that gave us permission to have a real conversation.

We were winning the opening and losing the sale.

AI is not a shortcut to trust

Here is what I had not factored in properly: in B2B sales, especially at higher price points, the relationship starts before the call. The outreach is not just logistics. It is the first expression of how you think, how you care, and what kind of partner you will be. When AI handles that at scale, prospects feel the difference even when they cannot name it.

Our team started having shorter discovery calls. Deals were taking longer to close. Referrals, which had been one of our strongest channels, dropped off for two quarters.

We had automated the part of the process that was actually doing the relational work. And we had no idea, because the top-of-funnel metrics looked healthy.

The decision

We pulled AI out of the outreach process entirely. Not out of the business. We kept it in research, in proposal drafting, in internal documentation, in briefing preparation. Areas where the output feeds into a person's thinking rather than replacing it.

The outreach went back to being written by people. Slower. More deliberate. With the rough edges that come from someone who genuinely looked at your LinkedIn, read your recent posts, and formed an actual opinion before reaching out.

In the following quarter, booked calls dropped by 30%. Closed deals went up by 40%.

The lesson I was actually learning

This was not a lesson about AI being bad. It was a lesson about knowing where AI creates value and where it quietly destroys it.

AI is brilliant at tasks where the output is the end product: a report, a summary, a proposal draft, a data analysis. It is risky in tasks where the output is a signal of character. Outreach is a signal of character. So is how you handle a complaint. So is how you respond to a referral.

In those places, the efficiency gain you get from automation is real. But the cost is invisible until the pipeline starts to feel slightly, inexplicably slower.

The question every founder needs to ask is not "where can AI help?" It is "where does the human signal matter more than the speed?"

Where does this leave you?

I am not suggesting you pull AI from your sales process. Your situation is different from ours. What I am suggesting is that you look at every place you have introduced AI and ask honestly: is this making us faster, or is it making us cheaper to feel like?

Because those are very different outcomes. And only one of them compounds over time.

What is one part of your business right now where AI is handling something that used to carry your character? And are you sure that is the right trade?

Recent Posts

By Dennis Kriel May 30, 2026
Most businesses struggle with AI because their leader hasn't made a clear decision about what to change. Dennis Kriel explains why AI adoption is a leadership problem, not a technology problem.
By Dennis Kriel May 22, 2026
Most founders want to simplify. Not one of them finds it easy. Here is why simplicity requires more discipline and harder decisions than adding complexity ever does.
By Dennis Kriel May 20, 2026
Most leaders stall at AI implementation not because of the technology but because of scope. Here's what real implementation looks like in practice, and how to pick the right place to start.
By Dennis Kriel May 4, 2026
The Memory Revolution: Why AI That Remembers Is a Different Game Altogether Persistent AI memory isn't a feature — it's a different paradigm. Here's what's changing, why it matters, and what to do about it in the next six months.
By Dennis Kriel April 30, 2026
The Agentic Era: Why Your Next Best Employee Might Not Be Human
April 16, 2026
Most businesses celebrate one AI win and then plateau. The Evolve step of the GUIDE Framework is where you build the internal habit of continuous improvement so your AI capability compounds over time.
April 14, 2026
Implementation gets AI started. Development makes it last. Learn why building capability in your people matters more than adding the next tool.
April 14, 2026
Most businesses implement AI once and stop. The Develop stage of the GUIDE framework is where implementation becomes lasting capability. Here is what that looks like in practice.
April 13, 2026
Most founders leave AI training inspired. Then nothing changes. Step 3 of the GUIDE framework is where inspiration must become a specific, measurable change — or quietly die.
April 10, 2026
Most founders think familiarity equals understanding. It doesn't. Step 2 of the GUIDE framework is where AI implementation either succeeds or quietly fails.
Show More