Liberate Launches Argus 1, the First Suite for Insurers Working With Personal AI Agents

Liberate today announced Argus 1, a comprehensive product suite designed to empower insurers to interact with personal AI agents. Launching as the flagship solution within the suite, AI Intercept provides rapid detection of personal AI bots during live phone interactions.

Personal AI agents such as Meta Muse and ChatGPT Dots reached consumers this fall, and they are already shopping for insurance. Most insurers can’t tell when the caller isn’t a person, making the AI agent a new kind of customer. In an October Liberate Survey of 1,000 U.S. policyholders, 21% said an AI assistant had requested an insurance quote on their behalf in the past month, and 54% said they would be likely to switch if their insurer refused to work with their AI assistant and a competitor would.

When an AI agent calls today, insurers have two bad options. They can tie up a customer service rep reading a script to a machine while a family with a burst pipe waits on hold. Or they can hang up and lose a real customer. Argus 1 is the third option.

“We’ve been planning for the rise of personal AI agents for a long time,” said Amrish Singh, co-founder and CEO of Liberate. “We’re building toward a future where a customer’s AI talks directly to their insurer’s AI and gets an answer in seconds, with no phone call at all. Argus 1 is how insurers get there.”

How AI Intercept spots a personal AI agent

AI Intercept answers the first question every call has to answer: who, or what, is on the line? In the opening seconds of a live call, it scores three kinds of signal:

  • How the caller sounds: the acoustic features of the voice
  • How they phrase things: the linguistic makeup of each sentence
  • The rhythm of the exchange: cadence and turn-taking, which differ between people and machines

When AI Intercept identifies a caller as an AI agent, it routes the call to a structured lane built for machines. There, the insurer’s core systems handle the request in a few turns without tying up a rep. Callers identified as human go straight to a person, with full context. When the model can’t tell, the call takes a third path that asks one direct question instead of guessing. In an offline test on 100 recorded calls, the detection model behind AI Intercept caught 96% of AI-agent calls (47 of 49).

From phone calls to direct connections

Detection is only the first stage. Argus 1 is built to follow the shift to AI-to-AI communication in three stages.

  • Today: voice. AI agents call in human language, and AI Intercept detects them and routes the call.
  • Next: structured exchange. A verified agent sends everything the insurer needs in a single request and gets one complete answer back, with no speech or transcription and a record on both sides.
  • Soon: direct connections. The customer’s agent connects straight to the insurer’s AI over a secure, governed channel, with no call at all.

“Voice is the bridge, not the destination,” said Singh. “Whichever protocol wins, someone still has to know who the agent works for, what it’s allowed to do and who signed off.”

Same rules for every caller

One set of rules applies across all three stages, built around disclosure, authorization and accountability. Liberate’s own agents already identify themselves as AI. Insurers set the rules once. A customer’s agent can handle routine service on its own, but it needs the customer’s approval to change coverage or payment and must hand any denial or dispute to a licensed person. Every exchange carries a full record of what was asked, what was offered and who approved it. Policyholders want the same accountability. In the survey, 80% said they want a real person at the insurer accountable when something goes wrong in an AI conversation.

Availability

Insurers, agencies and MGAs can join the Argus 1 waitlist today by visiting www.liberate.ai/argus-1

About Liberate

Liberate builds insurance-native AI agents for property and casualty insurance. Its agents answer calls, texts and emails, resolve sales, service and claims requests end to end and write every action back into core systems like Guidewire, Duck Creek, Snapsheet, Applied Epic and more. A Supervisor Layer governs each interaction and escalates to a human when needed. More than 70 carriers and brokers run 10M+ transactions a month on Liberate. Insurance runs on Liberate. liberate.ai

Survey methodology

On behalf of Liberate, Pollfish conducted an online survey on Oct. 17, 2026, of 1,000 U.S. adults, ages 18 and over, who hold car, homeowners or renters insurance. Results are weighted by age and gender using weights supplied by Pollfish, and all figures are weighted percentages. The maximum sampling error is approximately ±3.1 percentage points at a 95% confidence level. The 21% figure is the share of all respondents who selected “Had it request a quote from an insurer for me” when asked what they had done with an AI assistant in the past month. The 54% figure combines respondents who answered “very likely” (17%) or “somewhat likely” (37%) to the question, “If your insurance company refused to work with your AI assistant but a competitor would, how likely would you be to switch?” The 80% figure combines the two responses that named “a real person at the insurer” as accountable (54% and 26%) when respondents were asked which view on AI disclosure and accountability came closest to their own.

Media gallery