Agentic Barbarians at the Carriers’ Gates

Personal AI agents and the future of P&C insurance distribution

AI agents can already gather quotes, compare coverage and help consumers switch. The routes carriers open, or close, in the next 12 months will shape how P&C insurance is shopped for years.

Collaborative agentic AI ecosystem diagram

In September 2026, Meta launched Muse.

Two weeks later it was the
No. 1 free app. OpenAI followed with Dots.

Early tests show these agents can already return real quotes from some carriers.

The window for carriers to shape agentic shopping is open now, and it is closing.

Insurance shopping is entering its third phase

For 15 to 20 years, digital insurance shopping has moved through a web phase and an LLM phase. The agentic phase is small today, but it could change more, faster, than either.

~92%

~7%

<1%

Web

LLM

Agentic

Web: branded advertising, search and human-led comparison

As consumers went online, direct carriers that mastered digital acquisition grew share, slowly and steadily. Paid search and comparison marketplaces grew around them. Most carriers opted out of pure price comparison over concerns about commoditization, adverse selection and inconsistent information.

Consumers are moving faster than their insurers

Phase 2, LLMs helping shoppers research insurance, is well underway. AI search tools are becoming a meaningful source of high-intent shoppers, and their influence is larger than the traffic numbers show.

~50%

of U.S. adults have used LLMs, including ~30% of auto and home insurance customers

32%

of auto insurance shoppers used AI tools at some point in their 2026 search

~19x

growth in insurance site visits from AI tools since late 2024, the highest share of any industry studied

~41%

of insurance searches show a Google AI-generated answer, counted as ordinary search traffic

Four capabilities carriers need for LLM-driven shopping

Each web-era tool has an AI-era counterpart. Open a row to see what it involves.

Content that LLMs can find, trust and cite when shoppers ask about insurance. High information density, original statistics, off-page mentions from reputable sources and technical accessibility are levers carriers often overlook.

Paid demand will not disappear. Google sells ads in AI Overviews and is expanding into AI Mode, and OpenAI opened its platform to advertisers in 2026. Carriers should be testing into these placements methodically.

An LLM app or connector that receives shoppers inside the conversation and moves them through the sales funnel. Liberty Mutual and Allstate already offer ChatGPT apps that quote through conversation instead of a traditional form.

Visibility into AI-driven visits, quote sessions and costs, including AI agents reviewing renewals. Google added AI reports to Search Console in June 2026, but carriers will need their own tools to measure how accurately they appear across platforms.

Personal AI agents arrived sooner than most expected

Think of them as digital executive assistants: give one a task and guardrails, and it gathers quotes, weighs options and acts, keeping a human in the loop for the final call. EverQuote tested Muse and Claude across policyholder profiles. The agents can shop, though not always well.

Close up of hands typing on laptop. Night work concept.

Asked for quotes from every available carrier, Muse found 24. It returned verified quotes from five.

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

5 verified quotes (2 large, 3 midsize carriers)

8 blocked by website or access barriers

10 required follow-up from an agent

1 had no policy available in California

Even the verified quotes showed materially different coverage limits and inconsistent detail on deductibles, discounts and bundle savings. Shoppers still need guidance to compare policies fully.

~$800

cheaper: the quote Muse found in an analyst’s test, though it missed key policy details

42%

of consumers who used AI to shop for a new policy in 2026 completed a purchase

1.3x

as likely to switch carriers: AI users compared with non-users

Three forces will shape how fast agentic shopping grows

Consumer behavior, AI platform decisions and, most importantly, carrier actions. Pick one to see what is at stake.

Consumer adoption will set the pace, and insurance demands more trust than most purchases

Early signs point to rapid growth. But to run the full shopping process, an agent needs sensitive personal and financial information, and the trust to choose suitable coverage. An unsuitable choice could leave a household exposed at claim time.

Many consumers may let agents research and gather quotes while keeping control of purchases and switches. Even in that role, agents make it easier to shop more often.

Platforms must work through regulation, compute costs and conflict with their own advertisers

~63M

tokens used in one EverQuote shopping experiment with Muse: roughly $65 to $650 in compute, depending on the model. Over the past 18 months, the cost of similar AI performance fell by a factor of 750.

Exhibit C · Legal and compliance questions for AI platforms

01 Producer licensing may apply when platforms recommend or sell insurance

02 Comparisons that misstate limits or deductibles risk violating unfair trade practice laws

03 Cash-back offers may be restricted by state anti-rebating laws

04 Policyholder data must be protected and consent requirements met

05 Credit-report access needs a permissible purpose and safeguards

06 Agents presented as advisers could face liability if consumers end up underinsured

Carrier actions matter most, and agents bring four new risks

Price-led adverse selection

When agents compare mainly on price, carriers that underprice a risk win outsized volume, and losses surface later.

Policy life expectancy compression

Perpetual re-shopping shortens tenure and leaves less time to recover acquisition costs. EverQuote calls it an existential threat.

Legal and regulatory exposure

An accurate quote can still lead to a poor outcome if an agent misstates coverage or omits a required disclosure.

Human agent channel conflict

If AI agents take share from insurance agents, carriers face hard choices about protecting a channel they depend on today.

At the carrier’s gate, every access point needs a decision

AI agents will take whatever route gets them a quote: carrier websites, agency sites, aggregator feeds, insurance agents and, with voice, contact centers. At each one, a carrier can raise a wall, build a bridge or open a curated passage. Choose an option to see it play out.

Quoting engine

Enterprise website

Agency websites

Aggregator feeds

Insurance agents

Contact center

Trusted intermediary

● Decision point

◆ Personal AI agent

Block AI agent access, and risk losing new business growth

Blocking can limit quote traffic and its costs. It will not stop existing policyholders from using agents to find alternatives at carriers that don’t block.

~6.5%

of its premium book a non-participating carrier could lose each year once AI agents handle 30% of shopping

Illustrative model: 47.2% of in-force auto policies shopped in the prior year, × 30% of shops handled by AI agents, × a 46% switch rate for AI users. Assumes no offsetting agent-sourced business and equal average premium per policy.

Build direct connections, and risk losing control of how products are compared

An approved interface lets agents request quotes straight from the quoting engine. It costs money to build, maintain and police, and agents that check repeatedly raise quoting and data costs without raising sales.

Cheap, direct quotes also let price-led agents find a carrier’s lowest offers again and again. Once a quote leaves, the agent decides how it is read: it may ignore service reputation, miss coverage differences or treat different deductibles as equal.

Shape the channel through a trusted intermediary

Route agent access through one or more intermediaries on terms each carrier sets, including which products are offered and how they are represented. The intermediary handles the technical, commercial and analytical work of agent access and accurate quoting.

Shopping carrier sites one by one is expensive today, which gives AI platforms a reason to prefer efficient, carrier-approved routes. That opening narrows as costs fall, and this path likely offers the fastest time to market for a broad base of carriers.

Five moves for carriers to make before the market sets the terms

Measure AI agent traffic in your quote flows

Set a clear policy for how AI agents can access your products

Structure product and coverage data for AI and intermediary channels

Track the incremental cost of AI-driven quote activity

Decide your channel posture at every access point

The full whitepaper sets out the evidence, the economics of AI-driven quoting and each option in detail.

Agentic Barbarians at the Carriers’ Gates

In September 2026, Meta launched Muse.

Two weeks later it was the
No. 1 free app. OpenAI followed with Dots.

Early tests show these agents can already return real quotes from some carriers.

The window for carriers to shape agentic shopping is open now, and it is closing.

Insurance shopping is entering its third phase

Web: branded advertising, search and human-led comparison

LLM: AI search and chat recommendations

Agentic: AI that shops on the consumer’s behalf

Consumers are moving faster than their insurers

Four capabilities carriers need for LLM-driven shopping

AEO: LLM-optimized content strategy

Paid AI placements

AI storefront

AI reporting and telemetry

Personal AI agents arrived sooner than most expected

Asked for quotes from every available carrier, Muse found 24. It returned verified quotes from five.

Three forces will shape how fast agentic shopping grows

Consumer adoption will set the pace, and insurance demands more trust than most purchases

Platforms must work through regulation, compute costs and conflict with their own advertisers

Carrier actions matter most, and agents bring four new risks

Price-led adverse selection

Policy life expectancy compression

Legal and regulatory exposure

Human agent channel conflict

At the carrier’s gate, every access point needs a decision

Block AI agent access, and risk losing new business growth

Build direct connections, and risk losing control of how products are compared

Shape the channel through a trusted intermediary

Five moves for carriers to make before the market sets the terms

In September 2026, Meta launched Muse.

Two weeks later it was the No. 1 free app. OpenAI followed with Dots.

Early tests show these agents can already return real quotes from some carriers.

The window for carriers to shape agentic shopping is open now, and it is closing.

Insurance shopping is entering its third phase

Web: branded advertising, search and human-led comparison

Consumers are moving faster than their insurers

Four capabilities carriers need for LLM-driven shopping

AEO: LLM-optimized content strategy

Paid AI placements

AI storefront

AI reporting and telemetry

Personal AI agents arrived sooner than most expected

Asked for quotes from every available carrier, Muse found 24. It returned verified quotes from five.

Three forces will shape how fast agentic shopping grows

Consumer adoption will set the pace, and insurance demands more trust than most purchases

At the carrier’s gate, every access point needs a decision

Block AI agent access, and risk losing new business growth

Five moves for carriers to make before the market sets the terms

Two weeks later it was the
No. 1 free app. OpenAI followed with Dots.