Panta's founder initially tried selling SaaS to insurance brokers but found they resisted change. The more lucrative strategy was to build the AI tools for internal use, become the broker themselves, and capture the massive efficiency gains directly, rather than trying to sell efficiency to change-averse incumbents.
True AI adoption isn't about tools; it's a fundamental shift in organizational design. Traditional companies operate like Roman legions with layers of management. AI-native businesses are flat, allowing a single leader to manage thousands of AI-augmented workers directly, eliminating middle management and radically increasing efficiency.
In legacy industries like insurance and law, incumbents often claim their advantage is 'relationships.' In reality, this is often a euphemism for high friction and annoyance in switching providers. Customers stick with subpar service not out of loyalty, but because the effort of moving (e.g., finding 40 documents) is too high.
In heavily regulated industries like insurance, large carriers must justify their rates to the government. Becoming significantly more efficient could lead to regulators forcing price cuts, thus reducing revenue. This creates a perverse incentive to maintain high operational costs and headcount to protect their pricing power.
The efficiency of AI has reached a tipping point where it is quicker to train a model on a new, specific process than it is to hire and train a human. Once taught, the AI performs the task perfectly and scales instantly, while every new human requires the same lengthy and imperfect training process.
While top traditional brokerages average ~$350k in revenue per employee, AI-native models can flip this. By using AI to eliminate most back-office roles and augment salespeople, companies like Panta can exceed $1M per employee annually. This recurring revenue model can achieve a better unit economy than even tech giants like Google.
AI will do to white-collar work what the industrial revolution did to artisans. Complex knowledge work is broken down into simple, repeatable tasks that AI orchestrates. Humans become replaceable cogs in the system, performing the simple grunt work that machines can't yet do, fitting into the machine's process rather than the other way around.