A lot of the AI tools we have right now aren’t really transformation.
They make you faster, but you’re still prompting, approving, checking, updating, and figuring out what happens next. The better version is almost the opposite. AI should remove steps. It should show up where people already work, take something meaningful off their plate, and get easier to trust the more you use it. That’s a much harder product problem than just building something impressive.
Most AI Still Makes You Do the Work
AI assistants suck. AI copilots suck. AI employees are where it’s at. Before the ChatGPT fan club loses its mind, I use an AI assistant every day. It makes me faster, but I still have to send the email and chase the follow-up. A copilot prepares the next move, then waits for me to approve it. Great. Another queue to get through.
An AI employee gets the job. It follows up, handles responses, updates the CRM, and asks for help when it needs it.
We surveyed 250 companies using this framework:
- 79% were at the assistant level
- 17% were at the copilot level
- 4% were at the AI employee level
Most companies are using AI to do the same work faster, then putting “AI transformation” in the board deck.
Guys, you bought ChatGPT.
The bigger ROI opportunity comes when AI takes work off your team’s plate. I don’t care how many prompts your team ran last month. Show me more meetings held, more customers helped, or more work completed without asking for another five hires.
More on the data here.
Signals Are Hot. Making Them Useful Is the Hard Part.
Signals are the hottest thing in sales right now. Finally. Anything to shut the ABM crowd up. A guy asked me the other day, “If everyone’s following the same signals now, isn’t it already too late to stand out?” Everybody’s talking about signals, but half don’t know how to find one and most wouldn’t know what to do with one if it landed in their lap. You’ve got time, my friend.
Luckily, I’ve cracked the code. You knew I wasn’t going to talk this much shit without bringing up my own product :) We just built an AI Employee that watches accounts, catches signals, does the research, and finds the right people. And here’s where we got really innovative: we made it easy for the rep.
Right in Slack or Teams. Here’s what happened, why it matters, and who to call. Need more? Ask it. Somehow we managed without another login or a six-part webinar on “Signal-Based Growth.”
Make the rep’s life easier, you idiots. That’s the code.
I Could Also Churn My Own Butter
“Just build it in Claude.” Dude, I run sales. I don’t want a second job babysitting somebody’s Minecraft project.
Meet Dex, my sales assistant at Atonom. I talk to him in Slack, but there’s a whole lot going on behind that window.
- He watches my accounts for signals and pings me when something’s worth a look.
- He works through a waterfall of contact databases to find the right person and their actual contact info.
- He’s hooked into our CRM, so he knows who we’ve talked to, what happened, and can update it when needed.
- He also knows our company and ICP. Kind of important if he’s going to tell me who to call.
So I can ask, “Who should I call, why now, and what do we already know?” Then get back to selling. I can already hear somebody: “Gabe, I could build that myself.” I could also churn my own butter, dude. I’m going to have it delivered :)
Here’s My Magic Quadrant
Congratulations on your Gartner Magic Quadrant placement. Nobody cares anymore.
Every SaaS company’s pitch: “AI is making the old categories obsolete.”
Every SaaS company’s announcement: “We won one. Please clap.”
Apparently the old categories are BS until they hand you a trophy. Somewhere along the way, SaaS convinced itself that every feature deserved a category. Every category deserved a platform. And every platform deserved a three-year contract.
Now you’re selling me an AI employee that works across those categories. Great. Show me what your customer no longer has to do themselves.
Here’s my Magic Quadrant:
- Does it do the job?
- How much software can I delete?
Put your dot there and I promise I’ll clap when I cancel something.
Your AI Shouldn’t Have to Pay Admission
Salesforce loves AI...until you buy it from someone else. Then apparently there’s a toll booth. Salesforce is preparing to charge Flex Credits when third-party AI agents interact with its platform through MCP or direct APIs. Hell of a way to welcome everyone to the “agentic enterprise.”
Think about this as a sales leader. You spend years begging reps to update the CRM. You buy tools, hire ops people, build workflows. Still chasing people for notes on Friday. Then you finally get an AI employee that updates the opportunity, logs the conversation, and keeps the records current. Your CRM vendor’s response? Another billing opportunity.
Your team entered the data. Your company pays for the platform. Now your AI needs to pay admission to work with it. Apparently “bring your own AI” means “bring your wallet.” I build AI employees at Atonom, so yes, I have a dog in this fight. Our customers want work done across their systems. They shouldn’t need a calculator every time an agent updates a damn field.
Salesforce, if your AI is better, win the business. Earn it. Taxing customers for choosing someone else is a hell of an admission.
I Want to See What Someone Actually Built
I've said A LOT about AI in sales. Getting results takes more trial and error than most people admit. We've put it to work inside our own team. We've tested workflows, cut tools that weren't helping, and learned where AI actually produces ROI.
On November 18th, we're hosting the Autonomous Sales Summit to talk about what it actually takes to get results.
I don't need another keynote telling me AI is going to transform sales. I want to see what someone built, and whether it made them any money.
Register here.
Who the Hell Is Actually Worth Following?
Who the hell is actually worth following for AI + Sales on LinkedIn?
Seriously.
Who’s cutting through the noise and saying something that matters about how AI is reshaping your GTM?
- Not sharing “10 Claude prompts that replace your sales team” when you’ve never managed one.
- Not letting Claude ghostwrite their “thought leadership.”
- Not saying “Comment AI” to unlock some BS Claude created in 12 seconds.
- Not selling 2018 SaaS with Claude taped to its ass.
I want names. Because everyone’s an AI + Sales expert now, including the guy who was selling crypto last week.
So let’s fix that.
Tag the people in your feed who actually make you stop scrolling. Let’s build a list of operators, thinkers, and GTM killers who are pushing the conversation forward.
Who you got?
Zoom Out
I think we’re getting close to the point where just having AI in the product stops being interesting.
There are already hundreds of tools that sound cool. The harder question is which ones are actually worth keeping.
Right now, we have the tools, but we’re still doing half the work around them. I don’t want another thing I have to manage. I want something I can turn on and stop thinking about.
That’s where I think this is going. The winners won’t make people rebuild the way they work around AI. They’ll quietly remove enough work that eventually you barely notice the technology is there.
People are starting to realize how much work they’re still doing for their AI.
The next question is which tools actually make that worth it.
