Polsia: AI Agent + Zero Employees = $10M Run Rate

Eric Ries: Why Anthropic Won and How To Build Incurruptible companies

Presented by Zapier: https://zapier.com/ Timestamps 00:00 - Intro 01:09 - Anthropic’s mission and early structure 03:00 - FTX, investor alignment, and mission protection 04:03 - The Long-Term Benefit Trust 06:09 - Why trust can become a business asset 07:30 - Is Anthropic winning because of trust or product? 09:09 - What happened to Google’s original culture 10:30 - Why Google missed the transformer opportunity 12:18 - Companies that stayed true to their ethos 12:54 - What makes an organization incorruptible 14:42 - Is the MVP still possible in the AI era? 15:18 - Why Eric is worried about vibe coding 16:12 - The danger of building artifacts you do not understand 17:51 - Why creators overvalue what they make 19:30 - AI-generated work and the illusion of quality 20:06 - Validated learning versus AI artifacts 21:36 - Craft matters more than typing code 22:21 - How Eric uses AI for writing 25:48 - Solve It and Eric’s AI-assisted writing workflow 28:39 - Why editing AI output inside the context matters 31:39 - How Eric uses research with AI 33:18 - Using AI to evaluate and improve writing 35:24 - Keeping the human in control of context 36:09 - Final thoughts on Incorruptible  Eric Ries, author of The Lean Startup, joins Andrew to talk about his new book Incorruptible and why some companies stay mission-driven while others slowly lose what made them valuable. They dig into Anthropic’s founding structure, why trust can become a business advantage, what Google’s AI story reveals about corporate drift, and why Eric thinks the vibe coding era could end badly if people use AI to replace skill instead of building it. The big idea: AI can make builders more powerful, but only if it strengthens human judgment, craft, and learning. The artifact is not the asset. The learning is. 👉 Join us: https://thenextnewthing.ai/

Polsia: AI Agent + Zero Employees = $10M Run Rate

Polsia claims any non-technical person can launch and run a startup using their AI agent platform. Bold claim. So I asked Ben Cera to show me the real data: revenue, churn, infrastructure costs, and whether the product is actually building businesses — or just generating AI slop. Presented by Zapier: https://zapier.com/ Resources: https://thenextnewthing.ai/l/how-ben-cera-builds-polsia Polsia: https://polsia.com Here's what we covered: -How Polsia hit a $10M annual run rate with no full-time staff -The 50% month-one churn problem — and why Ben says it's not a disaster -An Anthropic bill that hit $1–1.5M/month, and how they're fixing it -Why the company name "Polsia" is "AI Slop" spelled backwards -The $30M funding round and what it's being used for -What it actually takes to build a zero-employee AI company in 2025 This is the hardest I've pushed a founder in a while. Watch until the end — his answer on whether 10% of companies making any money is a success will either inspire you or infuriate you. Chapters: 00:00 - Intro 00:59 - Polsia's $10M run rate 02:18 - Churn and who Polsia is for 06:27 - What needs to change in Polsia 10:12 - Is Polsia creating AI slop? 15:00 - Are users making money? 18:18 - Showcasing companies built on Polsia 21:36 - Why Polsia is AI slop backwards 25:30 - The $1.5M Anthropic bill 28:03 - Agent infrastructure partners 29:42 - How users bypassed email limits 35:15 - Zapier sponsor segment 36:27 - Cold outreach, spam, and guardrails 40:48 - AI-generated ads and Meta 43:39 - How much revenue comes from ads 44:33 - Andrew asks Ben to show Stripe 47:06 - Polsia coin and crypto scams 48:36 - Running Polsia with zero employees 52:12 - Ben's vision for agentic AI 54:54 - Closing 👉 Join us: https://thenextnewthing.ai/

I earned $500k when AI replaced my managers

👉 Link to resources: https://thenextnewthing.ai/l/chandler-lovable-adds-sales 👉 Chandler Bolt (X): https://x.com/chandler_bolt Presented by Zapier: https://zapier.com/ Timestamps 00:00 - Intro 00:27 - The AI sales management hub 01:12 - How every sales call gets graded 02:15 - AI feedback for sales reps 03:00 - Building the hub in Lovable 04:12 - Letting the team update scripts and rubrics 05:15 - Replacing managers with AI 06:36 - Automating call reviews and quality control 08:24 - Where human leadership still matters 09:27 - The librarian for sales stories 10:21 - Managers vs. leaders 11:42 - How much time AI saves 12:54 - What happens to manager roles 14:06 - Closing Chandler Bolt runs an eight-figure company, and his team used Lovable to build an AI sales management hub that helped add half a million dollars in sales last month. In this episode of The Next New Thing, Chandler shows how selfpublishing.com is using AI to grade every sales call, give reps detailed feedback, surface improvement opportunities, and turn call reviews into a repeatable system. Instead of managers reviewing a few calls per week, the AI hub reviews every call against a rubric, summarizes what happened, and gives specific coaching on what the rep can do better. Chandler and Andrew also talk about what this means for the future of management. The big shift is that AI can take over the repetitive parts of management, like quality control, call reviews, scorecards, and accountability, while human leaders focus on coaching, encouragement, strategy, and recruiting.

“I make $4.5 million implementing AI”

Presented by Zapier: https://zapier.com/ 👉 Link to resources: https://thenextnewthing.ai/l/jon-45million-ai-playbook ⏱ Timestamps 00:00 From $400 to $4.5M ARR 00:18 Hitting $1M solo with AI 01:57 The $105K dev quote that changed everything 02:15 Discovering Replit 03:18 From app idea to business opportunity 05:33 Having AI interview you to refine a business 07:21 The first business model 10:03 The first real customer and first $15K contract 13:12 Teaching practical AI, not just theory 16:48 Jon’s playbook for starting an AI business 19:03 How Zapier fits into the stack 21:18 Why the .org brand worked 23:15 Using GEO to get found in search and ChatGPT 27:27 Building recurring revenue with a fractional CAIO model 31:39 The move from services to software 33:54 What Jenna does for client businesses 36:18 Why this could become a billion-dollar business 39:00 How Jon thinks about pricing 41:33 Brand, trust, and distribution 43:39 Favorite tools: Grok, Replit, Midjourney, NemoClaw 47:51 Margins, growth, and reaching $4.5M ARR 49:12 How long this opportunity will last He started with $400, built the business himself with AI, hit $1M solo, and is now at $4.5M ARR. In this episode of The Next New Thing, Andrew Warner talks with Jon Cheney about the exact playbook he used to turn AI tools into a high-ticket recurring-revenue business.

Convos: Instant OpenClaw on your phone

Presented by Zapier: https://zapier.com/ 👉 Priority Launch List: https://thenextnewthing.ai/l/shane-priority-launch-list 👉 Shane Mac (X): https://x.com/ShaneMac 👉 Shane Mac (LinkedIn): https://www.linkedin.com/in/shanemacsays/ 👉 XMTP: https://xmtp.org/ ⏱ Timestamps 00:00 Launch AI agents on your phone 00:09 Copy any app with a prompt or screenshot 00:18 Creating an agent inside Convos 00:36 Agents provisioned with tools automatically 00:45 OpenClaw vs Hermes agents 00:54 What makes something an “agent” 01:21 Limited rollout and waitlist access 01:30 Turning a screenshot into an app 02:06 Demo: calorie tracking agent 02:33 From app → personalized AI coach 03:18 Training agents with personal data 03:54 Building a fully customized fitness assistant 04:30 Why agents get better over time 05:06 Backing from Andreessen Horowitz + USV 05:15 Coordinating group events with agents 06:00 Replacing chaotic group chats 06:45 Agent managing RSVPs, timing, logistics 07:21 Real-time updates and humor in chat 08:06 Monitoring content with “Radar” agents 09:00 Tracking writers, artists, and updates 09:45 Daily summaries across the internet 10:30 Personalized alerts and insights 10:57 Relationship + life coordination agent 11:24 Daily plans, reservations, and logistics 12:09 Combining multiple tools into one system 12:18 Product rollout and waitlist strategy 12:54 Future integrations (Notion, calendars, etc.) 13:21 Why messaging becomes the main interface 13:39 Agents talking to other agents 14:06 Privacy and coordination between agents What if your apps weren’t apps anymore—but agents you talk to inside a chat? In this episode of The Next New Thing, Andrew Warner sits down with Shane Mac to explore Convos, a new platform where you can launch AI agents directly on your phone—and have them act like full apps inside a conversation. Instead of downloading tools, you create agents by describing what you want. They get provisioned with email, phone numbers, browsing, and memory—then join your chats like participants. From there, you can clone apps, coordinate events, track information across the internet, or even build personalized systems that evolve over time. Shane demos how a simple screenshot can turn into a working app, how agents can act as assistants inside group chats, and how they can coordinate with other agents without exposing your personal data. The bigger idea: the interface is shifting from apps to conversations—and agents become the layer that connects everything you do.