AI agents run real browsers, pace themselves like people, and reason through your underwriting questions. ForMotiv tells you in-session whether a person, an AI assistant, or an autonomous agent is completing your application, before the policy binds.
Natural hesitation, corrections, recall
Person present, assistant filling fields
No human in the loop
Detection method
Scripted bots
AI agents
Headless browser checks
Caught
Real browser
Speed and velocity rules
Caught
Human-paced
CAPTCHA
Mostly caught
Solved
Device fingerprinting
Often caught
Rotated per session
Behavioral Intelligence
Caught
Caught
An agent can copy how fast a person types. It can’t copy how a person remembers their own life: pausing on a prior claim, fixing a typo in their street name, going back to check a date. ForMotiv models are trained on genuine human behavior across hundreds of millions of insurance sessions, so the gap shows up even when the surface looks human.
Agents that bind policies at scale with synthetic or stolen identities, with no plan to keep them.
Repeated quote-and-adjust runs that learn which answers reach the lowest rate or skip review.
Agents that start applications only to pull back the data your prefill vendors return.
One operator running many agents to sell fake or manipulated policies to real customers.
Real customers who hand the form to an assistant. Not fraud, but the answers weren’t entered by the applicant.
Agents that quote across carriers for a shopper. Low bind intent, high prefill and third-party data cost.
00:00
Real browser, clean device, no CAPTCHA trouble
00:14
Human pace, but no corrections at all
01:02
Answered with none of the pause people show when they’re remembering
01:40
Uniform timing on every field, no going back
02:05
Autonomous agent, routed to review
Blocking every non-human session means turning away real customers who use AI assistants. ForMotiv classifies each session and returns the evidence through the real-time API, so your team sets the response for each type of session and each line of business.
Keep verified humans on the fast path.
Ask an assisted applicant to confirm their key answers themselves.
Hold back expensive prefill and third-party data from low-intent agent sessions.
Send autonomous sessions to manual review or end them before bind.
Session classification returns through the real-time API in under 50 milliseconds on average. Use it to keep humans on the fast path, confirm assisted applicants, hold back prefill calls, or stop agents before bind.
Use the ForMotiv Console to measure how much agent traffic reaches each channel and Agent, group sessions run by the same operator, and review policies already on the books.
Agent patterns seen at one carrier strengthen detection at every carrier.
Signals and Models return in under 50 milliseconds on average.
See the Features behind each classification. No black box.
We capture behaviors, not identities. GDPR, CCPA, and PIPEDA compliant.
We’ll show you what agentic AI sessions look like in your own application data.
