How Insurance Agencies Use AI in 2026
Where AI fits inside US insurance agencies across the four operational categories: new business and sales, policy servicing, finance and commissions, and compliance. Operator-grade, AMS-aware.
AI in insurance agencies works through four operational categories (new business and sales, policy servicing, finance and commissions, and compliance), each containing specific, repeatable workflows where applied AI reduces manual effort or recovers lost revenue.
An operations guide organized by the four workflow categories that drive every US insurance agency.
By George Kaldelis, founder of Navicade. Fifteen years inside Allianz Greece running insurance operations before building Navicade to design, deploy, and operate applied AI workflows for US insurance agencies and EU/UK regulated trading firms.
This article is for agency owners and COOs who have moved past the experimentation phase and want a structured answer to a simple question: where, specifically, does AI fit inside an insurance agency, and which workflows are worth touching first. It is organized around the four operational categories that BPO sourcing firms, ACORD data standards, and every major AMS platform have built around for decades. We name the platforms, name the workflows, name the numbers, and cite the regulators. No futurism, no hype.
Why the AI conversation in insurance agencies keeps missing the point
The state of AI adoption in US insurance agencies is a wide gap between planning and deployment, and the gap exists because most agency-facing AI content is written by AI vendors rather than by operators. By early 2026, two-thirds of US insurance agencies plan to increase AI use in the next twelve months. Only 8% have AI embedded in daily workflows. The Big “I” ACT 2026 Tech Trends Report, Liberty Mutual’s 2025 Independent Agents at Work study, and Nationwide’s Agency Forward research all surface the same pattern: planning vastly outpaces deployment. The reason is rarely the technology. The reason is that most AI content for agencies is written by AI vendors, for AI vendors, organized around what AI tools can do instead of how an agency actually works.
The Big “I” ACT 2026 Tech Trends Report found that 38% of independent agencies say they are very likely to increase AI investment, another 30% are somewhat likely. But only 8% report AI embedded in daily workflows. 31% are not using AI at all. 55% lack written AI policies. The Liberty Mutual 2025 study of 1,242 agency leaders and team members puts the number at 1 in 3 employees using AI for work in the past year, with only 12% of agencies having a defined usage policy. Nationwide’s Agency Forward research reports that 24% of principals have committed to AI investment, with 48% planning to within the next year.
The pattern is consistent across studies. Intent is high. Action is low. The gap is what this article exists to close.
Most existing content for agency owners researching AI either lists AI tools (the vendor blog format) or summarizes AI categories (the consultancy explainer format). Neither answers the question agency owners actually have, which is: inside my specific agency, with my specific AMS and my specific book of business, where would AI fit and what would it change. That answer requires starting with how an insurance agency actually works, then asking where in that work AI applies. Not the other way around.
The four workflow categories that drive every insurance agency
Every US insurance agency, whether an independent agency, a BGA, an FMO, or a P&C retail shop, runs four categories of operational work. New business and sales covers lead intake through binding. Policy servicing and client service covers the daily work of keeping policies and clients current. Finance and commissions covers the money side: carrier statements, reconciliation, producer payouts. Compliance, licensing, and reporting covers everything regulators care about. This is the canonical operations taxonomy that BPO sourcing firms, ACORD data standards, and AMS platforms have organized around for decades. It is also where AI fits, workflow by workflow, inside each category.
The reason the four-category map matters is that AI does not apply to “an insurance agency” as a whole. It applies to specific workflows. Carrier commission reconciliation is a workflow. COI issuance is a workflow. Renewal management is a workflow. Each one has a defined input, a defined output, and a measurable cost when it goes wrong. Mapping AI to workflow categories instead of to the agency at large is how operators actually think about it. The IIABA Best Practices Study, the ACT Tech Trends Report, and every AMS vendor product roadmap organize around these categories because they reflect how the work is actually structured.
This is also the structure of the Navicade Insurance Agency Capability Map, the named taxonomy that organizes every Navicade Audit and Build Plan. For BGAs and FMOs specifically, the four categories apply with vocabulary shifts that LIMRA-NAILBA’s Inside the Intermediary research documents (downline, contracting, hierarchy, block of business), covered in the Finance and Commissions section below.
Here is the operational map.
Table 1: The four workflow categories in an insurance agency
| Category | Representative workflows | Where AI fits today | What is still manual |
|---|---|---|---|
| New Business and Sales | Lead intake, quote preparation and comparison, MGA submission processing, ACORD form population | Submission processing, cross-sell scoring, lead intake voice agents | Multi-carrier comparative quote selection, complex commercial new business |
| Policy Servicing and Client Service | COI issuance, endorsements, renewal management, claims intake | Email drafting, account summarization, pre-call account briefs | COI issuance and certificate holder workflows, multi-touch renewal outreach |
| Finance and Commissions | Carrier commission reconciliation, IB and producer commission splits, missing commission detection, finance reporting | Carrier statement matching, commission reconciliation in Applied Epic and AMS360 | IB commission split calculation, override management, variance investigation |
| Compliance, Licensing, and Reporting | Producer license tracking, E&O documentation, regulatory reporting, audit trail maintenance | Coverage research, producer licensing anomaly detection | Regulatory report generation, audit narrative, state DOI filings |
The rest of this article walks through each category, what AI actually does in it today, what is still manual, and where US agencies, BGAs, and FMOs are seeing measurable results. We will name the AMS platforms, name the workflows, name the agencies that have published results, and name the regulators every implementation has to respect.
New Business and Sales: where AI earns its first ROI
New business and sales is where most agencies see AI deliver measurable ROI first. The reason is volume: every quote, every submission, every ACORD form is a repeatable workflow with a defined input and a defined output. Vendor AI has moved fastest here. Applied Book Builder for cross-sell intelligence, Vertafore Velocity AI’s Portal Launcher Agent for MGA submissions. But the highest-ROI automation usually sits in two workflows: quote preparation and comparison, and submission processing for commercial and wholesale lines.
Quote preparation and comparison
The Ivans 2025 Agency-Carrier Connectivity Trends survey of 700+ industry professionals found that 72% of agents cite commercial submission automation as the top area for greater automation from carriers. Real-time risk appetite information ranked #1 as a factor in carrier selection. The pain is documented at the agency level: producers and account managers spend hours moving the same client data between carrier portals, rate engines, and the AMS.
Cedar Risk Management, a 40-year-old HawkSoft agency, uses ChatGPT, Claude, Grok, and Manus to compare quotes against bound policies and identify coverage gaps. Tanya Horst, who runs the implementation, reports 30 to 50% faster quote-to-proposal workflows and is rolling the process out to account managers and CSRs. The detail that matters: every output still gets reviewed before it goes to the client. As Tanya puts it, “we always review everything to make sure we’re not missing anything and nothing’s going to hit our E&O”. The AI compresses the work. The operator owns the result.
MGA and wholesale submission processing
For commercial and wholesale lines, the submission workflow is where the most manual effort lives. The same client data gets rekeyed into carrier portals, MGA submission forms, and the AMS account record. Vertafore’s Velocity AI Platform shipped a Portal Launcher Agent that automates navigation and form-filling on MGA carrier portals, reading application data from AMS360 and mapping it to carrier portal fields automatically. Vertafore reports 98% time saved on portal navigation. The Submission Processing Agent, in early adopter availability, reports 87% accuracy on automated intake.
Applied Systems’ Book Builder takes a different angle, enriching account profiles using public data sources to identify cross-sell opportunities and surface renewal risk. Stockman Insurance, a 65-employee 15-location agency in Montana, selected Applied Book Builder + Applied Recon in early 2026 to surface cross-sell intelligence and streamline commission reconciliation. Marcy Parks, Director of Operations, framed it as “smarter ways to manage increasing volumes of data and reduce the amount of manual work.”
McCurdy Group, a HawkSoft agency, uses Google NotebookLM to extract policy information from commercial submissions and generate coverage recommendations automatically. Dennis McCurdy, the founder, reports staff using it for policy questions without searching documents. “I find it a tremendous timesaver. My commercial lines staff is using it now.”
Lead intake and qualification, and ACORD form population, sit lower in the stack but still meaningful. InsuredMine’s Voice Agent (IVA) handles missed calls and new lead inquiries 24/7, qualifying intent and routing to the appropriate producer. ACORD form population is generally bundled into the submission processing workflows above rather than treated as a separate AI surface.
Policy Servicing and Client Service: the silent time drain
Policy servicing is the daily grind of an insurance agency and the workflow category that consumes the most CSR hours. Certificate of insurance issuance, endorsements, renewals, claims intake. All of it lives here. Vendor AI coverage in this category is uneven. AMS-native email drafting and account summarization have shipped at Vertafore, Applied Epic, and EZLynx. But three of the highest-volume workflows (COI issuance, endorsement processing, and multi-touch renewal outreach) have no native AI coverage at any major AMS platform as of mid-2026.
Certificate of insurance issuance
COI processing is the universal P&C agency time-sink. Industry research puts the volume at 3 to 12 CSR hours daily in a typical mid-size commercial lines agency, with 40% of weekly COI volume hitting Thursday afternoon to Friday morning. No major AMS vendor has shipped AI for COI generation. The workflow is still entirely manual: pull the request, identify the certificate holder, generate the form, deliver, log the activity.
This is one of the clearest gaps where independent AI deployment can outperform vendor roadmaps. HawkSoft partnered with Cara in September 2025, and HH Insurance, an early customer, reported over 50% reduction in routine reviews using Cara’s Servicing feature, which covers coverage comparisons, ACORD form filling, phone and email triage, and COI generation. The agencies on platforms that lack native COI AI (HawkSoft, NowCerts, AgencyBloc, QQCatalyst) have two paths: wait for vendor roadmaps, or compose AI into the workflow now.
Renewal management
Renewals are the second-highest volume workflow in policy servicing and the most operationally seasonal. Chris Paradiso, agency owner and Quandri customer, framed the state of it: “Renewals are harder than ever, and they are the key to running a profitable agency.” Quandri ships renewal intelligence for HawkSoft, AMS360, and Applied Epic, covering term-over-term comparison, requoting, and prioritization. Vertafore’s Email Agent in early adopter availability drafts renewal communications with 80% time saved and 98% accuracy claimed.
The harder workflow is the multi-touch renewal outreach sequence. Drafting an individual email is one task. Orchestrating a five-touch sequence across 200 renewing accounts in a single month is another. No AMS vendor has shipped AI that coordinates the full outreach orchestration. Agencies that automate this layer report measurable retention impact. Texas Trailblazer Insurance, an EZLynx customer, implemented EVA Email Content Assist for client communications. Suzanne Sansom, the president, reports measurable improvements in retention.
Endorsement processing consumes 30 to 40% of CSR daily workload in most agencies. Claims intake is FNOL on the agency side, distinct from carrier FNOL. Neither has shipped AI from any major AMS vendor.
Finance and Commissions: the high-error, high-stakes category
Finance and commissions is the workflow category where errors are most expensive and where AI has moved most aggressively in the past 18 months. Applied Recon reconciles carrier commission statements directly inside Applied Epic. Vertafore’s Reconciliation Agent reports 90% time saved and 94% accuracy across 200+ carriers. AgencyBloc’s Commissions+, though rule-based automation rather than AI, saves JTS Financial 90 hours per month and $33,000 per year. But for BGAs, FMOs, and any agency with multi-tier producer agreements, the harder workflow is downstream: IB and producer commission splits across hierarchy levels, override calculations, and chargeback handling.
Carrier commission reconciliation
This is where vendor AI has shipped most concrete capability. Applied Recon is built directly into Applied Epic, accepting carrier statements in any format and reconciling transactions using six-signal matching. Applied claims it is the industry’s only AI-powered reconciliation solution built natively into an AMS. Vertafore’s Reconciliation Agent on AMS360, in early adopter availability, reports 90% time saved, 94% accuracy, and support for 200+ carriers.
For agencies on platforms without native reconciliation AI, AgencyBloc’s Commissions+ is the high-performing alternative for L&H and benefits agencies. Note: AgencyBloc’s solution is rule-based automation, not AI. The agency configures column mappings per carrier, then Commissions+ handles CSV import, automated splits, and missing commission detection. JTS Financial, a 40-employee Little Rock benefits agency, reports 75% reduction in commission processing time, 90 hours per month reclaimed, and $33,000 per year saved. The commission manager previously worked weekends with her husband’s help to process monthly statements; now it is closed in normal hours.
The pattern across cases is consistent. Verlee B. at SIMA Benefits Consulting Group: “I finished commissions in record time this month, 5 hours, down from 5 days with our previous CRM.” Karen M Good reports 20 hours per week saved on commission processing, freeing her to focus on licensing and contracting for 140+ producers. Matt T. at Central Ohio Group Insurance: first-year savings on commission processing plus recovering missing commission payments exceeded the subscription cost.
IB and producer commission splits (BGA and FMO operations)
For BGAs, FMOs, and independent marketing organizations, reconciliation is only the first step. Once carrier transactions are matched, the producer commission split has to be calculated across the hierarchy. Downline producers, upline contracts, override tiers, chargebacks on lapsed policies. None of this is covered by carrier-level reconciliation AI.
No major vendor has shipped AI for IB commission split calculation or override management. The work remains rule-based configuration plus manual exception handling. The LIMRA-NAILBA Inside the Intermediary 4.0 study of 60+ BGAs and IMOs found that over 80% of intermediaries name expanding financial professional networks and increasing current product sales as top priorities. The operational mechanism for both is producer compensation accuracy. The NAILBA Technology Committee has surfaced repeatedly that BGAs believe they are systematically underpaid on their block of business across the carrier ecosystem of 20+ portals, 20+ statement formats, and 20+ contracting workflows. The multiplier effect is the real pain.
This is the workflow category where independent AI deployment can fill a gap that no AMS vendor has roadmapped. The agencies that solve it solve a problem the market has ignored.
Missing commission detection is partially covered by AgencyBloc Commissions+ at the statement level. Detecting missed commissions across carriers over time, including trend detection, is unaddressed by vendor AI. Finance reporting (variance explanation, board narrative) is similarly unaddressed.
Compliance, Licensing, and Reporting: the liability category
Compliance is the workflow category most agency owners avoid thinking about until something breaks. It is also where regulatory exposure compounds: NAIC Model Bulletins on AI use in insurance have been adopted by 24 states plus DC as of March 2025. New York DFS Circular Letter No. 7 (July 2024) holds insurers, and by extension their agencies, accountable for AI vendor outputs even when the agency does not directly run the AI model. Several states have added their own AI testing and disclosure requirements on top of the NAIC model. The agencies that implement AI without compliance grounding inherit the regulatory risk.
E&O documentation and gap identification
Every AI implementation in an insurance agency runs into E&O at some point. The fear is universal across operator interviews: a coverage recommendation drafted by AI is wrong, the client suffers a loss, the agency holds the license. Ron Kiefer at RPS frames the operating principle: “AI is new, it’s obviously still developing. It’s pretty important that in these early days of AI that there is oversight, we need human eyes on the output.” James Redeker at Swiss Re recommends: “Do the two side by side for the first few months. Don’t turn it over to AI until you’ve got it dialed in.”
Applied Book Builder includes coverage gap identification as part of its risk intelligence layer, surfacing accounts where E&O exposure is elevated. Vertafore’s Insurance Expert Agent, integrated with ReferenceConnect, provides natural-language search across 75+ trusted insurance publishers and 1M+ documents, returning cited answers to coverage questions. The agencies that have implemented these report that the value is not the AI’s recommendation itself but the documentation trail it generates. Every coverage decision is paired with the AI-surfaced research that informed it. That trail is the E&O defense.
The discipline that matters is what Tanya Horst at Cedar Risk Management described: review everything before it leaves the agency. “We always review everything to make sure we’re not missing anything and nothing’s going to hit our E&O.” Annette Ardler at Swiss Re extends it to vendor contracts: “Make sure you’re reading your vendor contract. That is not something carriers can help with because that’s legal advice, so agents would have to review it with their own counsel.”
The other discipline most agencies underestimate is shadow AI. When CSRs and producers use ChatGPT or other tools without agency oversight, the outputs never enter the audit trail. That creates regulatory exposure no agency can defend in an examination, because the documentation does not exist. A written AI usage policy plus a defined approved-tools list is how shadow AI gets brought into the audit trail.
Regulatory reporting and state filings
Producer license tracking, CE compliance, state DOI filings, NAIC reporting. None of this has shipped AI at any major AMS vendor with the exception of Vertafore Sircon’s anomaly detection on producer licensing workflows. State DOI filings remain manual. Regulatory report generation remains manual. Audit narrative generation is unaddressed.
For agencies operating in multiple states, the compliance overhead scales linearly with state count. NAIC bulletin adoption status varies by state, with New York, California, Colorado, and Connecticut having moved fastest. Multi-state agencies should track adoption state by state and align their AI usage policies to the strictest applicable bulletin.
The honest framing: compliance AI is the workflow category where vendor coverage is weakest, where the regulatory stakes are highest, and where most agencies have not yet built a deliberate AI approach. Audit trail maintenance is the foundation. Documentation discipline is the moat.
How AI fits around your AMS (and where it doesn’t)
AI for insurance agencies does not replace your AMS. It composes across it. Applied Epic, AMS360, EZLynx, HawkSoft, AgencyBloc. These are the systems of record where carriers, policies, accounts, and commissions live. Vendor AI agents operate inside their own product boundaries: Applied Recon reconciles inside Epic, EVA drafts inside EZLynx, Vertafore Velocity inside AMS360. The work that crosses systems (carrier portal to AMS to accounting to commission split to renewal communication) is where the AI layer delivers the largest ROI because no single vendor agent owns the full workflow.
Table 2: AMS native AI coverage matrix (as of May 2026)
| AMS Platform | Reconciliation AI | Email drafting AI | Renewal AI | Native AI status |
|---|---|---|---|---|
| Applied Epic | Applied Recon (shipped, native) | AI email summarization | Indirect via Book Builder | Strongest native AI coverage |
| AMS360 (Vertafore) | Velocity Reconciliation Agent (Early Adopter) | Velocity Email Agent (Early Adopter) | Velocity in development | Strongest agent roster, mostly invite-only |
| EZLynx | None | EVA Email Content Assist (shipped, bundled) | Patented workflow, not AI | One shipped feature, bundled pricing |
| HawkSoft | None | None | Via Quandri partnership | Zero native AI |
| AgencyBloc | Commissions+ (rule-based, not AI) | None | None | Zero AI, strong rule-based automation |
| NowCerts | None | None | None | Zero native AI |
| QQCatalyst (Vertafore) | None | None | None | Marketing-only AI positioning |
| InsuredMine (overlay) | None | Help Me Write, Communication Summarization | None | 7 AI features as AMS overlay |
Three of the seven major AMS platforms have zero native AI coverage. Vertafore’s most powerful agents (Reconciliation, Email, Submission Processing) are in Early Adopter, which is invitation-only, not general release. Applied Book Builder is a separate paid product, not bundled with Applied Epic. For agencies on HawkSoft, NowCerts, AgencyBloc, or QQCatalyst, AI is not on the AMS roadmap in any short-term sense.
This is where the operations question matters more than the tooling question. The agencies seeing the largest results are not buying AI tools and dropping them on top of the AMS. They are composing workflows that cross the AMS, the carrier portal, the accounting system, and the communication layer, with AI handling the steps in the middle that used to live in spreadsheets and email threads.
Where to start depends on the agency type. P&C commercial lines agencies in the 5 to 20 staff range typically see the largest first ROI on COI issuance and renewal processing, because volume is high and the workflows are repeatable. P&C personal lines agencies typically start with quote comparison and renewal reminders. BGAs, FMOs, and health and life agencies see the largest first ROI on commission reconciliation and producer contracting, because the carrier-side ecosystem multiplier is the dominant cost. Multi-line agencies should audit by category, identify the one workflow with highest cost times repeatability, and start there.
What you do not need: a dedicated AI team. What you do need: an honest map of where your operational cost lives, a clear sequence for which workflow to address first, and a partner who designs, deploys, and operates the system without handing you another tool to manage.
Frequently asked questions
Which AI workflow should an insurance agency implement first?
The right first workflow is the one where your agency’s cost and frequency intersect highest. For P&C commercial lines agencies, that is typically COI issuance or renewal management. For BGAs and FMOs, it is carrier commission reconciliation. For benefits agencies, it is commission processing and producer contracting. The decision is agency-specific, not industry-generic.
Does AI for insurance agencies require replacing my AMS?
No, AI for insurance agencies composes across your existing AMS, whether you run Applied Epic, AMS360, EZLynx, HawkSoft, AgencyBloc, or another platform. Vendor AI features (Applied Recon, EVA, Vertafore Velocity) operate inside their own AMS. Independent AI deployment composes workflows that cross the AMS, the carrier portal, the accounting system, and the communication layer. The AMS stays the system of record. Nothing gets ripped out.
What are the E&O risks of using AI in insurance agency workflows?
E&O exposure scales with how AI output is used. AI-drafted documents that are reviewed by a producer before going to the client carry the same E&O profile as documents drafted by a CSR and reviewed by a producer. AI outputs that go directly to clients without review carry elevated risk. The discipline that protects the agency is the same as the discipline that protects any error-prone workflow: human review on outbound documents, documentation of the AI-generated source material, and a written AI usage policy. NYDFS Circular Letter No. 7 holds the licensee accountable for AI vendor outputs regardless of who runs the model.
How long does it take to see ROI from AI in an insurance agency?
For a single workflow implementation, agencies typically see first measurable results within 30 to 60 days of deployment and ROI clearance within 90 days. The fastest documented results are in commission reconciliation, where Applied Recon and Vertafore Reconciliation Agent customers report time savings inside the first month. JTS Financial saved 90 hours per month and $33,000 per year on AgencyBloc Commissions+. Cedar Risk Management reports 30 to 50% faster quote workflows after a few weeks of use. The timeline shortens significantly when the implementation is operated rather than installed.
Will AI replace insurance agents?
No. AI in insurance agencies replaces specific repeatable workflow steps, not the agent. The work that remains squarely on the agent side is judgment under uncertainty: coverage recommendations for non-standard risk, client relationship work, claims advocacy when carriers push back, hard-market remarketing. Those workflows depend on context, trust, and discretion that AI does not provide. The agencies that thrive in 2026 are the ones that use AI to compress the repeatable workflows and free their producers and account managers for the judgment work that earns the renewal and the referral.
Where this goes next
If you are looking at those four categories and wondering which workflow inside YOUR agency would actually move the needle, that is what an AI Operations Audit answers. Seven days. Three named deliverables: a Revenue Leak Map, an AI Opportunity Map, and a Build Plan. Built to your agency’s specific bottleneck, not a template.
Or see how Navicade builds for US insurance agencies →
George Kaldelis is the founder of Navicade. He spent 15+ years inside Allianz Greece running insurance operations before building Navicade to design, deploy, and operate applied AI workflows for US insurance agencies and EU/UK regulated trading firms. Connect on LinkedIn.
This article describes how AI is currently deployed in US insurance agencies and is intended as an operational reference for agency owners, COOs, and prospective clients. It is not legal, regulatory, or compliance advice. Implementation of any AI workflow in an insurance agency should be reviewed against your agency’s E&O carrier requirements, state DOI regulations, NAIC Model Bulletin guidance where applicable, and any vendor contracts governing AI tool usage.