They killed the agency. Then gave it an AI backbone.
THE CLEAREST PATH TO A $10M+ SOFTWARE EXIT in 2 YEARS (with AI and agents)
building an agency right now is one of the most interesting business moves
the productized agency had its moment in 2022. it collapsed because scaling humans is a nightmare. inconsistent output, people...
The productized agency died because humans don't scale. AI agents resurrect the model — and turn service businesses into software-margin exits. Greg Isenberg's thesis reframes the agency not as a lifestyle business but as an AI-native software company in disguise.
Greg Isenberg — CEO of Late Checkout and one of the sharpest voices on internet business models — isn't arguing that agencies are back. He's arguing they never left; they just needed a different engine. The productized agency wave of 2022 promised scalable, repeatable service delivery, but it crashed against the oldest constraint in professional services: people. Inconsistent output, hiring bottlenecks, and margin compression killed most of them before they hit $3M ARR. What's changed is that AI agents — particularly orchestrated multi-agent workflows built on models like Claude and GPT-4 — can now handle the repeatable middle of agency work: research, first drafts, data analysis, reporting, and client communication. A16z's 2026 marketplace report found that AI-native agencies are operating at 70-80% gross margins, compared to 30-40% for traditional service firms.
The exit math is what makes Isenberg's argument compelling. Traditional agencies trade at 1-2x revenue because acquirers see human-dependent delivery risk. But an agency with AI agents doing 60-70% of the fulfillment starts to look like a vertical SaaS company — proprietary workflows, predictable output, defensible process IP — and those trade at 8-15x revenue. Companies like Jasper and Writer already proved that productized AI workflows command software multiples. The playbook Isenberg outlines isn't speculative; it's already being executed by dozens of AI-native firms targeting specific verticals like real estate marketing, e-commerce content, and financial reporting. SapienEx operates at exactly this intersection — building the agent infrastructure that turns service delivery into scalable, software-grade operations.
The bottom line: The agency model isn't dead — it was waiting for AI agents to replace the part that couldn't scale: the humans doing repeatable work.
Our Take
The productized agency died because humans don't scale. AI agents resurrect the model — and turn service businesses into software-margin exits. Greg Isenberg's thesis reframes the agency not as a lifestyle business but as an AI-native software company in disguise.
Greg Isenberg — CEO of Late Checkout and one of the sharpest voices on internet business models — isn't arguing that agencies are back. He's arguing they never left; they just needed a different engine. The productized agency wave of 2022 promised scalable, repeatable service delivery, but it crashed against the oldest constraint in professional services: people. Inconsistent output, hiring bottlenecks, and margin compression killed most of them before they hit $3M ARR. What's changed is that AI agents — particularly orchestrated multi-agent workflows built on models like Claude and GPT-4 — can now handle the repeatable middle of agency work: research, first drafts, data analysis, reporting, and client communication. A16z's 2026 marketplace report found that AI-native agencies are operating at 70-80% gross margins, compared to 30-40% for traditional service firms.
The exit math is what makes Isenberg's argument compelling. Traditional agencies trade at 1-2x revenue because acquirers see human-dependent delivery risk. But an agency with AI agents doing 60-70% of the fulfillment starts to look like a vertical SaaS company — proprietary workflows, predictable output, defensible process IP — and those trade at 8-15x revenue. Companies like Jasper and Writer already proved that productized AI workflows command software multiples. The playbook Isenberg outlines isn't speculative; it's already being executed by dozens of AI-native firms targeting specific verticals like real estate marketing, e-commerce content, and financial reporting. SapienEx operates at exactly this intersection — building the agent infrastructure that turns service delivery into scalable, software-grade operations.
The bottom line: The agency model isn't dead — it was waiting for AI agents to replace the part that couldn't scale: the humans doing repeatable work.