How to Actually Implement AI in Sales (Without Creating Chaos or Wasting Time) - Part 1

2 days ago
4 min read

Everyone in sales is “doing AI” right now.

t’s a checkbox. Teams are trying tools. Reps are using ChatGPT here and there. Leaders are talking about it. Someone on your team probably sent a “top 10 prompts” doc last week. Everyone nodded. Then went back to doing things the same way.

But here’s the reality:

Using AI is not the same as implementing AI.

And that’s where most teams are stuck.

I see a lot of sales organizations experimenting with AI… but very few are seeing meaningful impact or ROI from it. Not because the technology isn’t powerful, it is, but because we’re treating it like software we already understand.

We’re used to SaaS tools. You log in. You click around. You figure it out.

AI doesn’t work like that. It’s conversational. It’s contextual. And it only performs as well as you communicate with it.

Here’s the disconnect:

AI doesn’t speak human. And most of us don’t speak AI.

So, what happens?

You ask it to “rewrite this” → it rewrites it.You ask it for a strategy → it gives you one.You ask it for ideas → it gives you something that sounds right.

But it’s guessing.

It’s optimizing for something that feels helpful, not something grounded in your actual business, your deals, your buyers.

If you asked a real person to build a sales strategy, what would they do? They’d ask questions, push back, and try to understand your context.

AI won’t do that, unless you force it to. That’s why most teams are “using AI” … but not getting much out of it. They’re experimenting. Not implementing.

And even when teams get slightly better at prompting, AI still sits outside the actual sales workflow. It’s something you open occasionally. Not something embedded into how work gets done.

Opening ChatGPT once in a while isn’t an AI strategy. It’s curiosity.

Real impact comes when AI is built into your process.

From Experimenting to Implementing

So how do you actually make that shift?

Start with your process. Not the tool. Before you test prompts or buy anything new, map out how your sales team actually works today. Not the version in your playbook. The real version.

What happens day to day?
Where does time really go?
Where do deals get stuck?
Where do reps slow down, skip steps, or just “wing it”?

Write it out. Step by step.

You can do this as a team, whiteboard, Miro, whatever. Then have reps do the same individually. You’ll usually see gaps between what leadership thinks is happening and what’s actually happening. That alone is useful.

Then run a simple traffic light exercise:

Red: This is a bottleneck. It’s not working. It’s slowing us down or hurting outcomes.
Yellow: It works… but it’s inconsistent or inefficient.
Green: This part is solid. Not a priority right now.

This forces a more honest conversation.

Because most teams jump straight to: “What should we use AI for?”

Yet a better question is: Where are we losing time, consistency, or momentum?

That’s where AI might help. And I say might on purpose.

Because AI isn’t always the answer. Sometimes the issue is process. Sometimes it’s training. Sometimes it’s that no one truly owns that step.

AI is one tool in the toolbox. A powerful one, yes. But still just one tool. Once you identify the red zones, now you can ask a better question: Is this actually a good use case for AI?

What Workflows Should You Start With?

The first question usually I get: “What should we use AI for first?”

The honest answer is: it depends.

Who you sell to
How your buyers buy
Whether your team is inbound, outbound, or a mix
Your deal size
Whether you’re focused on new business or growing existing accounts

All of that shapes your sales process. So, use your traffic light exercise as your starting point.

Begin with red, then look at yellow.

Where most teams go sideways?

Not using AI in a way that improves the outcome. Right now, most teams are just scratching the surface. For example, every team has call recording software. You get transcripts, summaries, and maybe action items.

Yet most teams stop there. But that one call transcript could be used across multiple parts of your workflow:

It could shape your follow-up
It could inform your next meeting
It could update your CRM properly
It could help you prepare for objections
It could be compared against past calls to spot patterns

And that’s before you layer in:

CRM data
emails with the account
internal notes
manager feedback
deal context

This is the shift. It’s about how you combine inputs and think through the process.

The simplest way to think about it: Input → Process → Output

Start with the output. What are you trying to improve?

Better meeting prep. Stronger follow ups. More consistent discovery calls.

Then define the inputs. What information should go into that?

Last map the process. What would a strong rep actually do manually?

Review notes. Look at past conversations. Talk to their manager. Think through the account.

Now translate that into how AI supports your process, that’s when the output improves. The goal is not just faster, but better, more aligned, more consistent outputs.

If It’s Not in the Workflow, It Doesn’t Count

This is where most AI efforts fall apart. If it’s not part of the workflow, it doesn’t count. If your team is using AI when they feel like it, you don’t have implementation. You have experimentation.

Let’s say a rep writes 100 emails a week, and AI helps them do it 20-30% faster. Sounds like a win.

But here’s the question: Did that really move the needle? Or did we just make a noncritical task slightly more efficient? This is where teams get misled. They chase small efficiency gains instead of improving the parts of the workflow that drive revenue.

Real implementation looks different. It means: