What’s up, it’s Zayd.

I’m a day late with this week’s newsletter. Was locked in working on launching a few new features. But anyways…

This week I wanted to cover a few downsides of using AI in your outbound motion. Obviously, I think what we’re able to accomplish with AI is insane - but there are a few downsides you should be aware of.

Where AI Falls Short

AI-led tools are becoming integral parts of the outbound motion for powerful companies like Miro, Front, etc. And while AI has come a long way - allowing us to expand the output beyond what we could’ve fathomed years ago - it still has a long way to go.

If you’re using al AI-led tool to power your outbound strategy, there are a few things you should still be aware of:

1. BS Personalization

I once got a message that said, “Hey Zayd. See you’re in NYC, have you tried Blue Bottle Coffee? It’s great. By the way, come join our webinar.”

Personalized, but not relevant.

Everyone in NYC knows about Blue Bottle - but it’s a useless point to enter the conversation with.

This happens less and less the more you train the AI you work with - and the more the company improves their product on the backend. But it isn’t always easy to find the most relevant incision point.

Common Words and Phrases

Messages with…

I noticed

I was impressed by

That resonates with me

I’m keen to hear your thoughts

…scream AI.

In the past generation of messaging templates, phrases like “I stumbled on your profile” and “The algorithm told me to reach out” were the dead giveaways.

Of course my AI-dar is probably more well-tuned than the average person - given that I work with AI-generated text 24/7, 365.

But eventually, everyone will be able to sniff it out like this. Because eventually everyone will be exposed to it to a certain degree and it’ll become super noticeable.

This also improves as you learn to train the AI better and as the program itself improves, but you’ve got to be careful. Sending a cold DM with obviously AI-written text can diminish the relationship with the prospect in 2 seconds.

2. Incorrect Information

It’s difficult to marry a social graph into prospect conversations. If I try to use my LinkedIn connections for an AI-generated message, there’s a big chance for a mess up.

AI won’t know if we met at a prior event or if we both like Manchester United - so it can’t use that in a prospecting message. This means it could reach out with a less relevant message.

A “came across your profile” message to someone you met at an event is a massive yikes - and a prime example of times when AI isn’t good for nurturing existing relationships.

It’s much better for new cold outreach and connections.

3. It Still Takes Time to Ramp Up

SDRs take 90 days to ramp up. It’s going to take work, time, and thought to teach these tools to do what you need it to.

In the past 10 years of SaaS, there’s been a massive emphasis on reducing onboarding friction.

But with AI, it’s the opposite.

The output is only as good as the input. You have to introduce friction into the onboarding so you can get the AI dialed to the point where it can create outputs you’re pleased with.

Expecting it to be out-of-the-box and ready is reductive and pointless. Expecting this out of any AI tool is unrealistic.