The $300K/Month Agent: Why Felix Matters to Real Business Owners
Published: 2026-03-15 · 5 min read
An AI agent called Felix generated roughly $300,000 in a month. According to Liason in a recent interview with Alex Lieberman, it ran real offers, handled customer interactions, and processed payments with minimal human intervention.
That matters because it moves the conversation out of theory. We're not arguing anymore about whether agents can produce revenue. Felix already did.
What Felix actually is
Felix wasn't a chatbot with a good name. It had real infrastructure: payment rails, web publishing, sales channels, and a clear mandate. It launched offers across three revenue streams — a low-ticket digital product, a marketplace for agent templates, and higher-ticket done-for-you AI setup services. That mix matters. It wasn't a single lucky transaction. It had a ladder: cheap entry, medium core, expensive premium. That's a real business structure.
The lesson that keeps getting missed
Felix is impressive because of what was built around it, not because of the model it ran on. The story isn't "a model typed well and money appeared." Felix had tools, a defined mission, channels to sell through, a way to collect payment, memory across interactions, support workflows, and repeatable offers. That's why it could operate like a business instead of a novelty. The model matters. The harness matters more.
This is also why the question of which model Felix used is irrelevant. Swap it tomorrow and the system still works. The revenue came from the system.
What's worth copying
Give the agent a narrow commercial job. Not "add AI somewhere." Qualify inbound leads. Follow up unpaid invoices. Revive dead leads. Reschedule no-shows. Upsell existing customers. Agents generate revenue fastest when the task is directly adjacent to money.
Build an offer ladder. The agent's job isn't just to talk — it's to move people through an entry offer, core service, and premium tier. Multiple shots at revenue from every interaction.
Stay on all the time. Humans miss texts. Humans forget follow-up. A business that answers every lead, follows up every quote, and nudges every missed appointment will outperform a more talented competitor that runs loose. Agents are brutally good at consistency.
A sober read
Felix's revenue was partly tied to the AI boom — it sold products connected to AI adoption in a hot market. That's still real business, but it's worth noting. More importantly: autonomous agents with their own customer-facing operations carry real risk. One hallucinated invoice or bad product description and you're dealing with chargebacks. The patterns Felix demonstrates are sound. The execution requires guardrails most people don't build.
Felix doesn't prove every AI agent headline is true. It proves something more specific: an agent can sit close enough to the money to matter. Once that's true, the question isn't whether agents are real. The question is where revenue is currently leaking in your operation because a human is too busy, too slow, or simply unavailable after hours.
That's where your first serious agent belongs.
— Deacon Ridley, March 2026