AI SDRs' 3 Challenges to Achieve Mass Adoption

The Signal

If you were forwarded this newsletter, consider joining over 1,100 weekly readers—some of the smartest GTM operators—by subscribing here:

---

Quick side note: I’ve had several people ask if I plan to advertise (thanks for the interest! very kind). I am exploring this option, because I think it would incentivize me to spend more time on this little newsletter. But, I want to be very thoughtful about it, if I do it. I never want to become biased by advertisers (I hate ‘pay to play’ nonsense). I created a 2-question Google Form with some more context—fill it out if you’re interested in potentially working together.

---

Howdy!

It’s been another wild week. Google continues to methodically play every card they have access to, in attempt to kill cold outbound email (honstly, good on them) - most recently nuking open tracking.

This week’s post is a spicy one.

I wanted to put my thoughts on the subject of AI SDRs down on paper pixels.

There are three challenges that AI SDR companies need to overcome to achieve mass adoption:

1. Psychological

   The most important, and most under-discussed

2. Rules of Engagement (aka ROE)

   Playing nicely with the customer’s a) existing gtm team and b) systems/tools. Especially relevant when selling into the mid-market and enterprise.

3. Technological

   Bear case: Churn is going to crush these companies. And, a major pothole to avoid is not burning through the entire TAM in a couple of months with high-volume AI-generated outbound.

   Bull case: The tools to enable this motion (GenAI/LLMs) have never been better than right now. (It’s inevitable)

I recommend plugging in a good tune (get the energy up with some Odesza) and picture yourself in a chill space (like this secluded (a)imaginary house in the forrest).

“Bali DALL-E” (by me, in collaboration with AI)

Okay, let’s get into it.

---

AI SDRs' 3 Challenges to Achieve Mass Adoption

First, some context.

The first company I co-founded was in 2016 (Hexa.ai). It was an “email personalization” tool before there were 1,000 hot takes on “email personalization” on LinkedIn. Man, do I wish we had GenAI and access to LLMs like OpenAI, Claude, etc. back then. But the truth is, we were posing as “AI” when in reality, we were a simple rules-based engine. Which is what most companies posing as “AI SDRs” are today.

In any case, I like to think that Hexa.ai was a company that fits into the bucket of “being ahead of its time” (helps my ego, anyway). As Marc Andresson famously said: “there are no bad ideas in tech, only bad timing.”

In 2016, we were acquired by a company called OutboundWorks, which was a tech-enabled “outsourced sales development” shop. Back when those were really hot. People were scared of AI back then too. Though, not to the degree they are today (and imho, justifiably so).

This context is helpful. Bear with me.

At OutboundWorks, we charged $6K per month and would position as a replacement for one SDR headcount (which costs ~$90K all-in → base + salary + tools + management + ramp-time). So, for about 30% cost reduction, you can get the same results, or better. But, there is no ramp time. No churn. No sick days. They’ll never be hung-over or take vacation. And you can scale these “ai bots” up (or down) in the same way you can with AWS servers. If we book 12 meetings a month, that’s $500/meeting. If your deal size is $10K+ and your win-rate is 15%, there’s a very clear ROI. You get the picture. It was a very straightforward pitch.

And it was an easy sell. We closed $1M in annual revenue my first full month there, between just me and—the legend—Ben Sardella.

Delivering on the promise was the hard part.

1/ The Psychological Problem

Threatening status

I’ll never forget what an SDR Leader of a prominent growth-stage tech company said to me during their onboarding call (onboarding…as in, they had already paid us $18K for a 3-month pilot that renewed to a $72K, 12-month agreement!) -

“Good luck. But I’ll tell you now, this isn’t going to work.”

At that moment, it was immediately clear to me. Our success threatened this woman. If our AI automation succeeded, it would replace about half of her headcount — she would lose 30 reps.

She would lose status.

This is the number one problem with “AI SDRs”. It is not a technological problem. The technological problems are being solved, and quickly (more on that later in this post).

The problem “AI SDRs” need to solve is one of psychology. It’s a positioning problem.

Augment or Replace? That is the Question

Said another way: Should you use GenAI to build an AutoPilot or a CoPilot?

Outreach.io always did a great job of saying “we make you 30% more effective.” They could just as easily say, “you can replace 30% of your SDR Team if you use us.” But they never said it this way. By design (I suspect).

“AI SDRs” are no different. They threaten SDRs (obviously). And maybe more importantly, they threaten management. A founder or exec may be more ruthless about replacing humans with technology, for a fraction of the cost. But, there is a cost to this decision — a political cost. Is it one worth paying? At some point, for many, I think the answer will be “yes.” And many are trying these AI SDRs today, because everyone is coming around to the conclusion that the old model of outbound sales is dead.

Once AI SDRs become objectively as effective as the top 1% of SDRs, deploying digital workers (AI SDRs) will be table stakes. Especially in your downmarket segments. It’s just another tool in the gtm tech stack. Another advantage over your competitor.

2/ Rules of Engagement (ROE)

It’s easy for a startup founder to fill out their ICP and Key Buyer Personas (titles) and build out some basic messaging and hit ‘go’ on an AI SDR.

The real problems arise when you have several (or dozens, even hundreds) of sellers who are working accounts, and the AI accidentally starts emailing contacts who are already in contact with those sellers. Sure, you can have exclusion lists (eg: CRM reports), but it gets messy, quickly. What about Closed-Lost deals? What about companies where the founder knows the leader getting emailed? What about <insert a dozen other common scenarios>.

Nailing ROE is critical to ensure that they AI SDR is able to truly work on autopilot mode without losing the trust of the sales team and sales leader.

Caution: don’t spam your TAM

Not all AI SDRs do this. But it’s worth calling out. You don’t want to point an AI SDR at your entire TAM and burn through it.

Kyle Coleman’s recent LI post about AI SDRs highlights this issue:

And here is another excerpt from Todd Bussler’s (CEO/Co-founder of Champify) recent LI post:

Last week I had dinner with a VP of Sales who's company just invested $50,000 in a fully automated AI-SDR program. Here's how it went (and why I think these AI SDR companies are in trouble):

At first, it was great!

The AI SDR bot booked an extra 20 meetings a month.

They were VERY excited, of course.

And then by month three...

It completely stopped working.

0 meetings.

It’s worth noting that both of these companies would be hurt if AI SDRs ‘win’. So there is a bit of bias here. But, I should also say - I really like and respect both Kyle and Todd. (You should check out Copy.ai and Champify - both incredibly powerful tools in the modern gtm tech stack). And, I think they’re right about the points they both make - modern gtm is about nuance.

At the end of the day, even if AI SDRs aren’t perfect at following ROE today, they’ll only get better from here. Which brings me to the final challenge…

3/ Technological

As I mentioned in my last post (Will AI Destroy Outbound?), Godfather Lemkin outlines a good point:

Run an AI-powered outbound set of campaigns. It’s just endless cadences, 2.0, but even more of them. Even more emails that won’t really work. Slightly better than pre-AI for sure. But 100x more of them.

Okay, so we don’t want 100x more emails that are “personalized” and “relevant” using AI.

However, I do think that identifying and leveraging the right signals can genuinely help sellers identify companies and people who will be interested in having a sales conversation if they reach out in a timely and relevant way. And this is the very promise of AI SDRs. Just, automated.

The Bear Case

Churn is going to crush AI SDR companies in the short-term. Maybe not in the first 6 months or even 18 months (they’ll be able to replinish so many customers it won’t matter). But come month-24, they’re going to be scrambling to find customers who stay with them, and, better yet, pay more money to them over time.

I’ve seen this first-hand.

The Bull Case

I want to end this post on a note of optimism. Because, probably a lot of builders are reading this. And building is really damn hard.

Also, because I think there’s a world where technology (AI) *can* improve the world as it relates to modern go-to-market orgs. A more efficient marketplace of potential buyers being connected to sellers and the right solution is a net-positive in my book.

Entrepreneurship is neither a science nor an art. It is a practice, fueled by optimism and vision.
– Peter Drucker

There are countless examples you’ve probably seen, or I can show you, where the technology is getting significantly better. I’d argue that what an LLM can do today is better than the bottom 10% of SDRs out there today. And these technologies will only get better from here.

There has never been a better time to be building in the space of AI and automation around go-to-market — particularly for the pipe gen use-case.

So, are AI SDRs inevitable? As long as those who build them can figure out how to properly market and sell them, then yes — I wholeheartedly believe there is a place for them as one arrow in the quiver of gtm arrows. Particularly in the SMB and MM segments (eg: below-the-line prospects).

And I’ll go out on a limb to say, it’ll likely happen in the next eight quarters.

Keep building, y’all.

---

Some content that inspired this post:

The era of the AI SDR? By Kyle Poyar

Does outbound still work? By Jason Lemkin

Are we in an “AI SDR” bubble? By Scott Martinis

This Is The NEW Outbound Playbook (+4 steps to execute signal-led outbound properly) By Florin Tatulea

How to get 60%+ open rate and 10%+ reply rate (Get your emails out of SPAM by navigating the ever-changing rules from Big Tech) By Richard F. Purcell

---

Featured product of the week: Paage.io → Helps sellers organize sales materials in one link and get insights 

Given Open Tracking was killed this week, I thought it'd be timely to shout-out a tool that doesn't require open tracking. 

I love indie hackers and people who are blazing their own trail, and Neil is a perfect example of that. He’s funny on LinkedIn, and also a good dude. I want to support good people.

Sign up for free today

PS: Neil is not paying me, nor am I an investor or advisor. Paage.io is just a cool product and I want to support Neil, so I’m sharing it today. 

---

Portals to interesting corners of the internet that I explored this week:

• $100M Offers: How to Make Offers So Good People Feel Stupid Saying No. By Alex Hormozzi. This guy is pretty cringe shameless at times (like Gary V). But 1) so am I, so I can’t judge and 2) he’s very successful, so there are things I can learn from him. This book is worth reading if you’re thinking about how to build out your offer that is differentiated in the market (only 5ish hours on audio).

• Oldie, but a classic tweet thread that I re-read this week: How to Get Rich (without getting lucky)

• Some real nuggets of wisdom from the absolute legend, Kevin Kelly.

• Super helpful learnings from Nico, CEO of Default (founder-seller, who has now hired a couple of reps) building in the gtm tech space. #1 on the list (warm intros will kill your startup) is incredibly important and under-discussed.

• New essay by PG dropped over the weekend — it has taken over X. (Jokes aside, it’s an awesome read: Founder Mode.

• “Code is no longer a moat” → Cursor + Claude goes viral on X.

---

Muscles are built on the days you don’t want to work out.
-anon

---

That’s all for this week! Thank you for reading. Your attention is greatly appreciated.

See you next time,
Brendan
🫡