The Sales Leaders Who Survive the AI Transition - and the Ones Who Don't

I'm going to tell you something that most senior sales leaders won't admit: I didn't see the last AI wave coming fast enough, and I was sitting inside it.

I co-founded an AI company in 2013. We called it "predictive analytics" because nobody knew what machine learning was yet. Half our sales pitch was explaining that no, we were not a fortune-telling service. We were building recommendation engines and predictive personalization for Fortune 500 clients before "AI" was something the marketing team was allowed to put in a pitch deck. We got acquired by Rise Interactive in 2015. I watched a dozen peer companies either collapse under their own hype or get absorbed into larger platforms that buried the technology in a menu no one clicked.

I know exactly what a technology transition looks like from the inside. And I'm watching the same movie play out again - except this time, the underlying technology is genuinely more powerful than anything we had in 2013. Which makes the stakes for sales leaders considerably higher.

Two Types of Leaders, One Outcome

Here's the taxonomy that matters. There are two types of sales leaders right now:

Architects understand that their job is to design systems, not fill pipelines. They're learning to build workflows, configure agents, and think about how work actually gets done at every stage of the revenue process. They're not coders. They don't need to be. But they can look at a 40-step SDR workflow and identify which 15 steps are automatable, which 10 need human judgment, and which 15 don't need to exist at all.

Order-takers are waiting for their RevOps person or their IT team to "implement something." They view AI tools as software procurement decisions, not operating model decisions. They'll get a subscription to three different AI platforms in the next 18 months, implement none of them properly, and then report to their board that "AI hasn't really moved the needle for us yet."

Every sales leader I talk to says they're "using AI." When I ask how, 90% say "we have a Gong license." That's like saying you're a chef because you own a microwave. I am not being harsh. I'm being specific because the distinction matters. Order-takers built fantastic careers in an era when scale was achieved by adding headcount and processes were fixed at annual kickoffs. That era is closing. The next era rewards people who can redesign the machine in motion.

What "Building" Actually Looks Like

I set up a Mac Mini as a local AI agent server. (It hasn't asked for a day off or updated its LinkedIn to "open to work.") This is not a story about me being technical - I want to be honest that I'm not a developer and haven't been since I was writing bad HTML in 1999. This is a story about what's now accessible to anyone willing to spend a weekend on it.

The agent I built does four things: It scrapes target account news every morning across 40 sources. It synthesizes the relevant signals for each account. It drafts a brief for each rep. It queues outreach suggestions tied to those briefs. The whole thing runs before 6 AM and produces 50 account briefs by 6:15. When I had SDRs doing this manually, a skilled rep could do maybe 15 accounts per week.

I built this without writing a single line of code. I used no-code workflow tools, an off-the-shelf language model, and about 12 hours of iteration across two weekends. The cost is roughly $40/month in compute and API calls.

That's what "building" means for a non-technical sales leader in 2026. Not deploying enterprise software. Not managing an IT project. Designing a workflow, assembling the pieces, and iterating until the output is useful. Buying a $200K revenue intelligence platform and not changing your process is like buying a Peloton and hanging laundry on it. Don't be the Peloton story.

The Skill That Actually Matters

The capability that separates architects from order-takers isn't coding. It's systems thinking. It's the ability to look at a process and decompose it into discrete steps, identify which steps require judgment and which are mechanical, and design the handoffs between humans and machines.

Ironically, this is exactly what the best sales leaders have always been good at. A great VP of Sales looks at their territory and thinks: what does a rep need to do to close a deal in this segment? What can I systematize? What has to be improvised? Where do managers add value and where are they just attending meetings? That's systems thinking. It's the same skill. The vocabulary just expanded.

What's new is that the output of good systems thinking can now be implemented in hours, not quarters. Before AI tools, a sales leader could design a perfect onboarding process and then wait six months for RevOps to build it in Salesforce. Now you can build a working prototype in a weekend and iterate the following week.

A Self-Assessment You Can Do Right Now

Write down your current SDR or AE workflow - every step from lead acquisition to meeting booked or deal closed. Every single step. Most leaders come up with 30–50 discrete steps when they do this honestly.

Now go through each step and ask two questions:

1. Does this step require a human to make a judgment call based on incomplete or ambiguous information?
2. Does this step primarily require processing, retrieval, or transformation of information that is already defined?

Steps that answer yes to question 1 stay human. Steps that answer yes to question 2 - and only question 2 - are candidates for automation right now, today, with tools that exist and are inexpensive.

In my experience, 35–45% of a typical sales workflow falls into category 2. That's not a small number. That's almost half the work. If you've never done this exercise and you're running a sales org of any size, you're making headcount decisions without understanding what work actually needs humans.

The Pattern From 2013

Here's what I saw at SimpleRelevance and at the AI companies around us during 2013–2015: the companies that failed weren't bad at technology. Many of them had impressive tech. They failed because they tried to sell AI as a product when buyers needed AI as a workflow integration.

The AI companies that won - the ones acquired at real valuations or that built sustainable businesses - did something different. They embedded their technology so deeply into a specific workflow that removing it felt like removing a load-bearing wall. They didn't sell "AI." They sold a specific outcome made possible by AI.

The same principle applies to sales leaders in 2026. You're not trying to implement "AI." You're trying to redesign specific workflows so that specific outcomes happen faster, more consistently, and with fewer errors. That's the frame of an architect, not an order-taker.

The Honest Prediction

The next 24 months will sort the field. The sales leaders who come out stronger will be the ones who got uncomfortable, tried things, broke things, and iterated (and the associated insomnia). They won't all be technical. They'll just be the ones who treated workflow design as a core leadership skill instead of an IT problem.

Every CRO I know has sat in a board meeting defending a number they knew was fiction, nodding confidently while their internal monologue screamed. The ones who also used that discomfort as a forcing function to build better systems - those are the architects.

The ones who don't survive won't be replaced by robots. They'll be replaced by younger leaders who grew up thinking about automation as a normal part of their job, the same way my generation grew up thinking about Excel as a normal tool.

The transition is happening. The question is which side of it you want to be on.

joepeck.ai is my attempt to build in public - the Deal Coach, the Forecast Truth Machine, the Autonomous SDR demo. Every tool there is something I built because I needed it, using the exact approach I described above. Take a look if you want to see what the architect's path looks like in practice.