Evidence period: Current customer page checked 2 October 2026; launch date from Aviva releaseChecked in this release: 2 Oct 2026
Keep in mindAn app launch does not establish autonomous buying, UK availability or lower acquisition costs.
What supports this view?
Full interpretation: Conversational discovery and quoting can change acquisition routes. Intermediaries may adapt, and customer ownership may shift toward agent platforms.
Established: The evidence separates estimates, personalised quotes and purchase handoffs.
Evidence to change the assessment: Reassess when consented agent journeys produce incremental bound policies, lower fully loaded acquisition costs and acceptable retention, risk mix and customer outcomes.
Sigo opens agent-based motor quote-and-buy and InsuranceMCP directory
Evidence item dated 2026-10-01 · Announced
Sigo reports motor insurance quoting and in-agent purchase with selected carrier partners in Texas, plus an open InsuranceMCP directory.
Limit: Company claims, not independently tested. The release says Muse approval is pending; the Muse customer page describes web or WhatsApp completion. Preserve this journey discrepancy and Texas-only scope.
Evidence period: 2026-10-01Source published: 1 Oct 2026Checked in this release: 4 Oct 2026
Sterling scales OpenDialog AI sales agent across 53 motor schemes after controlled conversion test
Evidence item dated 2026-10-01 · Announced
OpenDialog reports that Sterling Insurance Services, part of Adrian Flux Group, deployed its Selma AI sales agent across all 53 Private Car schemes and then Short Term Car. In a controlled A/B test running January–August 2026, customers using the AI journey showed a 31.4% relative lift in quote-to-purchase conversion and a 9.3% relative lift in add-on attachment. Conversion lift reached 39–41% versus control in July and August. Selma detects likely quote abandonment and answers using the customer’s live quote, policy wording and IPID through Sterling’s API.
Limit: The primary evidence is a vendor case study and company press release, not independently audited. The source does not disclose sample size, absolute conversion rates, confidence intervals, customer cohort mix, premium or claims quality, complaint outcomes, cancellation rates, vulnerability outcomes, or whether add-on attachment improved customer value. The 31.4% figure is the January–August relative lift; 39–41% relates only to July and August and should not be described as the full-period result. The deployment is an on-site sales agent inside Sterling’s controlled journey, not an external personal agent shopping across marketplaces.
AI assistants can help people discover insurance and obtain prices. The important question is how far each journey actually goes, and who completes the sale.
Established: The evidence separates estimates, personalised quotes and purchase handoffs.
Developing: Delegated authority and transaction interfaces may support more complete journeys.
Still uncertain: An app launch does not establish autonomous buying, UK availability or lower acquisition costs.
Tuio / Lassie · Europe · 2026-09-09 Discovery: Reported · Estimate: Not specified · Quote: Reported · Purchase in agent: Not evidenced · Autonomous renewal: Not evidenced
Insurify · US · 2026-08-18 Discovery: Reported · Estimate: Superseded · Quote: Reported · Purchase in agent: Carrier-dependent · Autonomous renewal: Not evidenced
Aviva · UK · 2026-06-01 Discovery: Reported · Estimate: Reported · Quote: On insurer site · Purchase in agent: Website handoff · Autonomous renewal: Not evidenced
Simply Business · UK · 2026-04-23 Discovery: Reported · Estimate: Reported · Quote: On broker site · Purchase in agent: Website handoff · Autonomous renewal: Not evidenced
What changed in the evidence?
Sigo opens agent-based motor quote-and-buy and InsuranceMCP directoryEvidence item dated 2026-10-01 · Announced
Sigo reports motor insurance quoting and in-agent purchase with selected carrier partners in Texas, plus an open InsuranceMCP directory.
Limit: Company claims, not independently tested. The release says Muse approval is pending; the Muse customer page describes web or WhatsApp completion. Preserve this journey discrepancy and Texas-only scope.
Evidence period: 2026-10-01Source published: 1 Oct 2026Checked in this release: 4 Oct 2026
Sterling scales OpenDialog AI sales agent across 53 motor schemes after controlled conversion testEvidence item dated 2026-10-01 · Announced
OpenDialog reports that Sterling Insurance Services, part of Adrian Flux Group, deployed its Selma AI sales agent across all 53 Private Car schemes and then Short Term Car. In a controlled A/B test running January–August 2026, customers using the AI journey showed a 31.4% relative lift in quote-to-purchase conversion and a 9.3% relative lift in add-on attachment. Conversion lift reached 39–41% versus control in July and August. Selma detects likely quote abandonment and answers using the customer’s live quote, policy wording and IPID through Sterling’s API.
Limit: The primary evidence is a vendor case study and company press release, not independently audited. The source does not disclose sample size, absolute conversion rates, confidence intervals, customer cohort mix, premium or claims quality, complaint outcomes, cancellation rates, vulnerability outcomes, or whether add-on attachment improved customer value. The 31.4% figure is the January–August relative lift; 39–41% relates only to July and August and should not be described as the full-period result. The deployment is an on-site sales agent inside Sterling’s controlled journey, not an external personal agent shopping across marketplaces.
Evidence period: 2026-10-01Source published: 1 Oct 2026Checked in this release: 9 Oct 2026
The Zebra connects Muse to personalised insurance comparison resultsEvidence item dated 2026-10-01 · Announced
The Zebra announces an agent-to-agent connector that lets Muse or Instinct extract policy details and declaration pages from user-authorised files, pass them to The Zebra’s data harness and take the consumer to personalised US comparison results. Users can also ask the assistant to enrol them in automatic rate tracking.
Limit: US marketplace and company announcement; live end-to-end performance was not independently tested. The consumer lands on The Zebra’s results page, carrier data is shared only after active selection, and the source does not establish autonomous binding, renewal switching, UK availability or platform endorsement.
Sigo opens agent-based motor quote-and-buy and InsuranceMCP directory
Announced
Sigo · Muse, MCP
Sigo reports motor insurance quoting and in-agent purchase with selected carrier partners in Texas, plus an open InsuranceMCP directory.
INSURANCE IMPLICATION · EDITORIAL
Shows an agency offering structured carrier access directly to consumer AI agents, with implications for UK motor distribution and quoting costs.
Evidence limits & next signal
Evidence limitCompany claims, not independently tested. The release says Muse approval is pending; the Muse customer page describes web or WhatsApp completion. Preserve this journey discrepancy and Texas-only scope.
Watch nextMuse approval
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
Sterling scales OpenDialog AI sales agent across 53 motor schemes after controlled conversion test
Announced
Adrian Flux · AI
OpenDialog reports that Sterling Insurance Services, part of Adrian Flux Group, deployed its Selma AI sales agent across all 53 Private Car schemes and then Short Term Car. In a controlled A/B test running January–August 2026, customers using the AI journey showed a 31.4% relative lift in quote-to-purchase conversion and a 9.3% relative lift in add-on attachment. Conversion lift reached 39–41% versus control in July and August. Selma detects likely quote abandonment and answers using the customer’s live quote, policy wording and IPID through Sterling’s API.
INSURANCE IMPLICATION · EDITORIAL
A UK broker deployment moves insurance AI from lead generation to regulated quote-to-bind intervention with measurable funnel economics. It suggests that governed, product-grounded agents may lower effective acquisition cost by rescuing already-paid-for traffic and may influence add-on mix, while making consumer-understanding, relevance, vulnerability, human-handoff and outcome monitoring central controls. This is material for AI Distribution and should, if approved, refresh the topic-state-v4 current-state signal and implications in the same governed release.
Evidence limits & next signal
Evidence limitThe primary evidence is a vendor case study and company press release, not independently audited. The source does not disclose sample size, absolute conversion rates, confidence intervals, customer cohort mix, premium or claims quality, complaint outcomes, cancellation rates, vulnerability outcomes, or whether add-on attachment improved customer value. The 31.4% figure is the January–August relative lift; 39–41% relates only to July and August and should not be described as the full-period result. The deployment is an on-site sales agent inside Sterling’s controlled journey, not an external personal agent shopping across marketplaces.
Watch nextWatch for primary evidence confirming availability, customer journey and market scope, and for changes to the stated limitations.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
The Zebra connects Muse to personalised insurance comparison results
Announced
Zebra · Muse
The Zebra announces an agent-to-agent connector that lets Muse or Instinct extract policy details and declaration pages from user-authorised files, pass them to The Zebra’s data harness and take the consumer to personalised US comparison results. Users can also ask the assistant to enrol them in automatic rate tracking.
INSURANCE IMPLICATION · EDITORIAL
Demonstrates a governed handoff between a personal agent and an insurance marketplace, including context preservation, selective carrier disclosure and recurring rate monitoring—important for customer ownership, validation reuse and quote-volume control.
Evidence limits & next signal
Evidence limitUS marketplace and company announcement; live end-to-end performance was not independently tested. The consumer lands on The Zebra’s results page, carrier data is shared only after active selection, and the source does not establish autonomous binding, renewal switching, UK availability or platform endorsement.
Watch nextWatch for primary evidence confirming availability, customer journey and market scope, and for changes to the stated limitations.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
Neptune opens flood-insurance shopping to Muse agents
Reported live
Neptune · Muse
Neptune says consumers can ask a personal Muse agent to explore its US flood-insurance options. The agent guides the user through available coverage while Neptune’s digital platform supplies rapid pricing and supports policy issuance.
INSURANCE IMPLICATION · EDITORIAL
Shows a specialist, non-comparison-site insurance product being made discoverable to a general-purpose personal agent, directly relevant to gateway design, agent-friendly product access and quote-cost economics.
Evidence limits & next signal
Evidence limitUS flood insurance and a company announcement. The source describes agent-guided shopping and Neptune’s existing digital quote/issuance infrastructure, but does not demonstrate that Muse independently completes a binding purchase or renewal, quantify agent-generated sales, or establish UK availability.
Watch nextWatch for primary evidence confirming availability, customer journey and market scope, and for changes to the stated limitations.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
Insurify moves from estimates to personalised quotes
Reported live
Insurify · ChatGPT
Insurify reports real-time motor quotes in 11 US states. Purchase proceeds through the insurer or Insurify’s agency, depending on the carrier.
INSURANCE IMPLICATION · EDITORIAL
Comparison platforms can participate inside agents. Test conversion and acquisition cost by source, then compare the risk mix.
Evidence limits & next signal
Evidence limitProvider-reported. Its 5.7% AI-discovery attribution among new H1 2026 customers is self-reported attribution, not market share or completed in-agent purchases.
Watch nextExpansion beyond the initial states and attributable bound policies, costs and retention.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
EverQuote adds an agentic distribution partnership
Announced
EverQuote · Waniwani · Cross-platform
EverQuote announces an exclusive commercial partnership and minority investment in Waniwani, with plans to integrate its offering for P&C carriers and agents.
INSURANCE IMPLICATION · EDITORIAL
Acquisition intermediaries may adapt their role around agent discovery and conversion. Distribution fees need not disappear.
Evidence limits & next signal
Evidence limitA commercial announcement does not establish rollout, carrier participation, customer volumes or cost savings.
The Mills Review examines changing consumer journeys, competition, market power and risks. Its recommendations include enabling foundations for agentic finance.
INSURANCE IMPLICATION · EDITORIAL
Track customer control, accountable decisions and platform dependence alongside the commercial opportunity.
Evidence limits & next signal
Evidence limitReview recommendations are not new binding rules. Its consumer survey spans retail financial services, not insurance buying alone.
Watch nextSubsequent FCA decisions, consultations and implementation dates.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
Aviva adds an indicative life quote to its ChatGPT journey, following home insurance. Customers continue to Aviva’s website to complete the application and buy.
INSURANCE IMPLICATION · EDITORIAL
Conversational entry can shorten the start of a journey while underwriting and purchase remain with the insurer.
Evidence limits & next signal
Evidence limitAviva explicitly says customers cannot buy its insurance directly within ChatGPT. A final price may require further information.
Watch nextIncremental completed sales, handoff drop-off and differences in applicant mix.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
FIDO announces agent authentication and payment standards work, drawing on Google’s Agent Payments Protocol and Mastercard’s Verifiable Intent.
INSURANCE IMPLICATION · EDITORIAL
Provable delegated authority could help separate permission to research from permission to transact.
Evidence limits & next signal
Evidence limitGeneral commerce infrastructure. It does not establish insurance suitability, policy binding authority or a live insurance purchase journey.
Watch nextFinal specifications, interoperable deployments and clear liability allocation.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
Small business insurance starts with a four-input estimate
Reported live
Simply Business · ChatGPT
Simply Business launches indicative pricing using trade, turnover, years in business and UK postcode. Customers complete the full quote and purchase on its website.
INSURANCE IMPLICATION · EDITORIAL
Commercial insurance can gain conversational entry points. Estimate-to-final-price movement and completion are useful measures.
Evidence limits & next signal
Evidence limitAn indicative estimate is not an underwritten final quote. No autonomous purchase or conversion uplift is established here.
Watch nextHandoff conversion, final-price divergence and coverage suitability.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
Google introduces a shared protocol for agent commerce
Announced
Google · retail partners · Gemini / UCP
Google introduces Universal Commerce Protocol for discovery, checkout and post-purchase interactions, initially describing eligible US retail use cases.
INSURANCE IMPLICATION · EDITORIAL
Shared transaction interfaces are relevant to insurance distribution architecture, but insurance still needs product-specific decisions and controls.
Evidence limits & next signal
Evidence limitRetail infrastructure signal. The announcement does not show UCP-based insurance quoting or policy sales.
Watch nextInsurance-specific integrations with documented consent and purchase responsibilities.
Sources accessed 9 Oct 2026. Interpretation is editorial; capabilities have not been independently tested.
Scope and supported claims for 40% more quotes turned into policies, live on schedule in 14 weeks - OpenDialog - AI agents built for regulated industries
Used for: Sterling scales OpenDialog AI sales agent across 53 motor schemes after controlled conversion test