
Hosted by Gabe Larsen · EN

Most AI agents are toys. They read your inbox. They summarize reports. They draft a few emails. Nice productivity boost. But saving someone eight minutes reading email is not a revolution. The real shift is AI doing full jobs. In this episode we break down what that actually looks like in the wild. One large healthcare company deployed a single autonomous Cloud Employee to handle Tier 1 through Tier 3 support calls across their product lines. Today it: Handles 2,600+ support calls per day Covers 65 different support skills Operates in 15 languages Logs 124 hours of talk time per day The average human support call used to take 18 minutes. The AI resolves them in about 3 minutes and 35 seconds. When you convert that back to human time, this system replaces more than 600 hours of labor every day. That is the difference most people are missing. One version of AI helps you work slightly faster. The other replaces the work entirely. Most of the market is still building AI assistants. We are focused on AI workers. Learn more about Cloud Employees at Atonomhttps://atonom.ai Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

Insight Partners co-founder Jerry Murdock recently argued that the “human buyer” could disappear sooner than people expect. Most people interpreted that as AI simply helping humans research software faster. But the bigger implication is something very different. What if the buyer itself isn’t human? For the past two decades SaaS companies built everything around human limitations: Dashboards to interpret data Menus and interfaces to navigate complexity Demos and sales calls to guide decisions Marketing funnels and nurture campaigns to influence buyers The entire SaaS ecosystem evolved around a biological decision maker. But in an agentic world, that changes. Instead of a human researching vendors, an AI agent could evaluate products, run tests, compare results, and choose the best option automatically. That means: No demos No product tours No nurture campaigns No traditional buying journey Just machine-to-machine evaluation. If that shift happens, the companies that survive will design products and infrastructure that autonomous agents can consume directly. Everyone else may spend the next few years improving interfaces that humans never use. Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

Everyone in business has experienced it. You approach someone at an event, extend an arm… and suddenly you’re in the dreaded half-hug. Arms everywhere. Nobody knows what’s happening. For years Gabe was firmly in the “hug guy” category. Then the pandemic introduced the safest diplomatic greeting humanity has ever created, the fist bump. Now we’re living in a strange transition period. Some people hug. Some people handshake. Some hover awkwardly waiting for the other person to decide. It’s basically a real-time social negotiation. In this episode Gabe shares his current strategy for avoiding the awkward half-hug and asks the real question: Where do you land in the great workplace greeting debate? Handshake Hug Fist bump Or the dangerous “wait and see” approach. Connect with Gabe Larsen LinkedIn: https://www.linkedin.com/in/gabelarsen Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

White-collar workers are telling themselves a lie about AI. The common narrative says AI will replace low-skill jobs first. Factory workers. Retail workers. Service workers. But recent research from Anthropic shows something very different. The jobs with the highest exposure to AI are overwhelmingly white collar, including: Analysts Marketers Recruiters Finance teams Consultants In other words, anyone whose job revolves around: thinking writing summarizing analyzing These are exactly the types of cognitive tasks large language models are already extremely good at. And before someone says the usual line… “AI will just augment us.” Sure. For the top operators. But historically, companies don’t adopt technology so everyone can keep their job. They adopt it so fewer people can produce dramatically more output. Which raises a bigger question: What happens when one strong operator with AI can do the work that used to require an entire team? Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

For twenty years SaaS had a powerful moat: migration pain. Switching systems meant risk, downtime, broken integrations, lost data, and internal political battles. Most companies stayed not because the tools were perfect, but because leaving felt impossible. That dynamic is starting to change. We recently removed Salesforce and built our own CRM in Lovable. It wasn’t ideological and it may not be permanent, but it highlighted something important. Most companies already have the people needed to build and maintain systems: Salesforce admins RevOps teams Integration consultants You’re already paying someone to manage and maintain the system. The only question is what they’re building. Are they maintaining someone else’s platform… or building leverage inside your own system? At Atonom, we’re building Cloud Employees, AI workers that run real workflows across tools and systems. As AI begins to execute more of the work inside organizations, the economics of software start to shift. Per-seat pricing makes less sense. Interfaces matter less. Control of systems and data matters more. This isn’t the death of SaaS. But it may be the end of passive SaaS, where companies outsource too much of their strategic thinking to vendors. If AI is going to run more of the workflows in your business, you may not want to rent the operating system those workflows depend on. Learn more about Cloud Employees at Atonom:https://atonom.ai Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

Everyone keeps saying AI is about to replace SDRs. But something interesting is happening. OpenAI, the company building some of the most advanced AI models in the world, is hiring a Head of Sales Development. If AI was going to wipe out SDRs entirely, you would expect the company building the technology to lead the way. Instead, they’re scaling the function. The real shift isn’t the role. It’s the work inside the role. AI will absolutely replace a lot of the repetitive motion SDRs handle today, including: List building Account research First-touch outreach Follow-ups Scheduling That’s the repetitive execution layer. But the function of building pipeline doesn’t disappear. It evolves. Someone still needs to: Design the outbound strategy Orchestrate humans and AI Decide where automation works and where it breaks Own pipeline creation The companies that win won’t be the ones hiring huge SDR teams. They’ll be the ones building pipeline machines, where AI runs the motion and humans run the strategy. That’s exactly what we’re digging into at the AI SDR Summit on April 9. No paid tickets. No paid sponsors. No paid speakers. Just operators sharing what’s actually working. https://atonom.ai/events/aisdr-summit?utm_source=gabe&utm_campaign=aisdr Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

Calling out SaaS sellers: your buyer enablement sucks. Most sales processes still look something like this: Discovery Demo More demos Contract Close But that’s not how buying actually works. After the demo ends, the real process begins. Your buyer now has to go sell your product internally. They need to convince finance, leadership, operations, and often multiple stakeholders that the investment is worth it. And this is where most SaaS sellers fail. They think they’ve “enabled the buyer” because they: Sent over the same generic slide deck they use for every prospect Forwarded a call recording that no one internally will ever watch Said “let me know if you need anything” and disappeared That’s not buyer enablement. That’s laziness. Real buyer enablement helps your champion win internally. Every internal purchase requires three things: Current State Help the buyer clearly explain what is broken today and why staying the same is risky. Future State Help them paint a compelling picture of how things improve with your solution. ROI Help them justify the investment with a clear return, even if the math requires some assumptions. The buyer isn’t just deciding whether they like your product. They’re trying to build a case that survives internal scrutiny. If you don’t help them do that, deals stall or die. Not because the product isn’t good. Because the seller didn’t enable the buyer to win. Great SaaS sellers don’t just demo features. They help their champions become heroes inside their organizations. https://atonom.ai/ Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

“Something Big Is Happening” went viral for a reason. 80M+ views. That’s ~1% of the world. People didn’t share it because it scared them. They shared it because it confirmed something they already felt. The February 2020 analogy matters. Flights were full. Offices were open. Life looked normal. But the curve had already bent. AI feels similar. This isn’t 3.1 vs 3.2. This is linear turning vertical. What used to require: • Back-and-forth iteration • Multiple handoffs • Days of refinement Now happens in: • One well-structured prompt • One feedback loop • One autonomous build cycle The first industry to feel it? Software engineering. For 20 years, engineering was the safest bet in the economy. Then AI went from: Autocomplete → Code suggestions → Full scaffolding → Autonomous builders In under two years. Engineering wasn’t targeted. It was simply closest to the blast radius. AI labs optimized for code because code builds AI. Once that loop worked, the capability expanded. Now the compression spreads: • Law • Finance • Consulting • Customer support • Revenue operations Anywhere work = language + logic + structured process. That’s why the article resonated. One industry already compressed. The rest are debating whether the shift is real. This isn’t panic territory. It’s attention territory. The curve has already bent. The only question is whether you see it while it’s happening — or after it hits your function. Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

Vinod Khosla recently said BPOs could disappear within five years. Most service leaders will dismiss that. They shouldn’t. Traditional BPO is built on: • Labor arbitrage • Headcount scale • Utilization math • Margin on human throughput That model worked when labor was the constraint. AI changes the constraint. If your business relies on: • 1,000 agents answering tier-one tickets • Teams updating CRMs • Manual invoice processing • Rules-based lead qualification • Repetitive back-office tasks You are operating a temporary data-processing layer. And AI eats temporary layers. The opportunity isn’t to shrink. It’s to redesign. The next-gen BPO likely looks like: • 50 high-skill operators • 5,000 AI workers • Outcome-based pricing • 24/7 execution • No training lag • No attrition This is no longer about cost per FTE. It’s about orchestration per outcome. Nearshore will still matter for complex, judgment-heavy workflows. Offshore will still matter for scale. But the dominant layer becomes “Smartshore”: BPOs that specialize in AI agent orchestration and Cloud Employee management. The fork in the road: Defend seats and optimize headcount math. Or rebuild around output and orchestration. Comfort vs inevitability. The market won’t reward comfort. Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/

Salesforce is charging for Agentforce three different ways: • $2 per conversation • $0.10 per action • $125+ per user per month One product. Three models. Why? Because SaaS was built on seats. But AI replaces seats. If your AI works, customers need fewer humans. If customers need fewer humans, they need fewer licenses. If you charge per seat, your best product eats your own revenue. That’s the revenue paradox every incumbent is staring at right now. So companies are experimenting in public. The PricingSaaS 500 tracked 1,800+ pricing changes last year across the top 500 B2B and AI companies. That’s 3.6 pricing shifts per company in a single year. Nobody has conviction yet. We’re watching three camps form: Per-seat incumbents trying to protect predictable ARR. Usage-based vendors aligning price to compute and API calls. Outcome-based challengers trying to tie price directly to value created. This is not a feature update cycle. It’s a structural shift. For twenty years, software was access. Now AI is output. And output doesn’t map cleanly to logins. When software becomes labor, you price it like labor. That’s the shift. The companies that figure out how to align price with output, capacity, and business impact will win. Everyone else is just adjusting knobs and hoping the spreadsheet holds. Get my weekly breakdown of AI, GTM, and Cloud Employees:https://atonom.ai/newsletter Ready to hire your first Cloud Employee?https://atonom.ai/