AllDigital Specialty automates 70% of insurance business with AI

AllDigital Specialty is trying to prove a simple but uncomfortable point for the rest of the insurance market: if you start from scratch, AI can do far more than sit on top of old workflows. The Princeton, New Jersey-based MGA says roughly 70% of its business is handled autonomously by AI systems, with 30% to 40% of incoming business receiving automated approval, another 30% declined by machine systems, and only the remaining 30% sent to human review.

That split matters because it shows where the real advantage comes from. CEO Athula Alwis and his co-founders built AllDigital around machine learning from day one, without legacy infrastructure, entrenched process debt or internal resistance to unwind. In practice, the company uses machine learning and agentic AI across submission intake, clearance and pre-analysis preparation, turning work that usually slows underwriting distribution into a tighter, faster pipeline. AllDigital frames its maturity in four stages, recommendation, assistance, execution and orchestration, and Alwis says the firm is now at stage three.

The scale is no longer theoretical. Industry coverage in 2024 said AllDigital had written more than $100 million in premium, and the company launched in 2021. That makes it a useful benchmark for P&C software strategists who want to know what an AI-native operating model looks like once it gets past pilot language and into actual market production. It also gives incumbent carriers a reality check: the gains AllDigital is posting come from architecture choices, not just from adding a chatbot or a scoring layer to a traditional stack.

Alwis brings more than 30 years in insurance, with past roles at Chubb, Willis Re, QBE and Munich Re. AllDigital says he also helped grow Freedom Specialty to more than $400 million in revenue within five years, a résumé that helps explain why the company has been comfortable moving fast in a regulated market. But the firm is also explicit that automation has not replaced governance. Human oversight, model controls and guardrails remain central, a stance that lines up with the National Association of Insurance Commissioners’ model bulletin on artificial intelligence systems, adopted on December 4, 2023.

The broader market context is shifting too. The NAIC says AI is already running across underwriting, pricing, claims and fraud detection, which means the question is no longer whether insurers will use it. The sharper question is whether they can rebuild around it. AllDigital, based at 155 Village Blvd., Suite 310 in Princeton and staffed at roughly 11 to 50 employees, is betting that a clean-sheet platform, backed by cloud infrastructure and human accountability, can do what legacy carriers still have trouble copying.