The Hartford's AI Strategy: How a Top-10 U.S. Carrier Is Winning Small Business With Conversation

TL;DR#

The Hartford Financial Services Group is using conversational AI to fix the single biggest problem in small commercial insurance: business owners cannot fill out a 40-field Business Owner Policy (BOP) application without an agent on the phone. As a top-10 U.S. small commercial carrier with roughly $5 billion in annual small business premium and more than 1.5 million small business customers, The Hartford has invested heavily in narrative-driven intake — business owners describe what they do in their own words, AI classifies the operation into NAICS and SIC codes, and an agent or appointed broker takes the hand-off pre-qualified. The shift matters because 60% of U.S. small businesses are uninsured or underinsured according to the Insurance Information Institute, and the dominant friction is not price — it is application complexity. Commercial insurance AI built around conversation, not forms, is how Hartford keeps an 11% small commercial market share defensible against InsurTech entrants like Next, Pie, Coalition, and Hiscox. This post unpacks Hartford's playbook, what bundling BOP, workers' comp, and cyber via conversation actually looks like, and what it signals for the $130 billion U.S. small commercial market.

What is The Hartford doing with AI in small business insurance?#

The Hartford is replacing static commercial application forms with conversational AI that lets a small business owner describe their business in plain language and produces a structured, classifiable, bindable submission on the back end. The carrier — founded in 1810 and headquartered in Hartford, Connecticut — has telegraphed this direction across its 2023, 2024, and 2025 investor days, where executives have repeatedly pointed to "small commercial digital" as the highest-ROI technology investment in the company. Hartford reports roughly $5 billion in annual small business premium, more than 1.5 million small commercial customers, and one of the broadest BOP appetites in the market — covering more than 1.3 million classes of business across retail, professional services, contractors, and specialty trades.

The strategic rationale is straightforward. Hartford's small commercial book is one of the most attractive in the industry — higher margins than mid-market, stickier than personal lines, and harder to disrupt than auto. But the front door — the new-business application — has not meaningfully changed in two decades. A typical BOP submission asks 30 to 60 questions, half of which depend on industry classification the applicant cannot self-determine. Conversation collapses that flow into something a coffee-shop owner or HVAC contractor can complete in five minutes on a phone. For background on how the broader category is moving, see the commercial insurance AI guide for brokers, MGAs, and carriers.

Why small business owners abandon commercial applications#

Small business owners abandon commercial applications because the language, time commitment, and classification complexity exceed what any non-insurance professional can navigate. The National Association of Insurance Commissioners (NAIC) tracks small commercial premium growth annually and consistently flags that "application friction" is the number-one cited reason agents lose new-business quotes. The Insurance Information Institute estimates that 40% of small businesses operate without business insurance — not because they don't want coverage, but because the path to it is too painful to complete unassisted.

Three specific failure modes drive abandonment:

• Language. A BOP application asks the owner to self-classify into NAICS or ISO class codes. "Are you a 'special trade contractor — electrical' or a 'building construction — general contractor'?" The owner doesn't know, guesses wrong, and the submission gets re-rated downstream — sometimes increasing the premium by 30%+.
• Time. Industry data suggests a typical small commercial application takes 22–35 minutes to complete on a desktop, and far longer on mobile. Most owners are filling this out between customers or after closing, not in a dedicated work session.
• Complexity. The application demands payroll, prior loss runs, square footage, year built, alarm/sprinkler specs, and certificate-of-insurance requirements from existing landlords or clients — most of which the owner has to look up.

Conversational intake fixes all three at once. The owner types or speaks a description ("I run a 1,200-square-foot pet grooming salon with two part-time employees and one mobile van"), AI extracts the classifiable entities, asks follow-up questions only where it has low confidence, and surfaces a clean structured submission to underwriting. This is the same pattern documented in the conversational intake guide for replacing forms with conversations, applied to commercial lines.

Inside The Hartford's small-business conversational intake#

Hartford's small-business intake operates as a three-layer system: narrative business description, AI-led classification, and broker or agent hand-off. The carrier has not published a public architecture diagram, but the customer-facing flow and the language Hartford executives have used in investor calls map cleanly to a now-standard conversational underwriting pattern.

Layer 1 — Narrative business description. Instead of opening with a NAICS dropdown, the flow opens with a single open-ended prompt: "Tell us what your business does." The owner answers in 1–3 sentences. AI then asks 4–8 targeted follow-up questions only where the description is ambiguous — for example, distinguishing "general contractor" from "remodeling contractor" by asking about the share of new-construction work. This is the same probing model behind Perspective AI's interviewer agent, and it is what makes conversational classification reliable.

Layer 2 — AI-led classification. Hartford's underwriting engine maps the narrative to a class code, looks up the appropriate BOP appetite, and produces an indicative premium range — all before a human ever touches the file. This is the layer where InsurTech-native competitors like Next Insurance and the AI-first SMB insurance playbook have set the bar; Hartford's response has been to fold the same automation into its existing carrier infrastructure rather than spin up a new brand.

Layer 3 — Broker or agent hand-off. Roughly 80% of Hartford's small commercial premium flows through independent agents and appointed brokers. The intake system does not try to bypass them — it pre-qualifies the risk so the agent's conversation is about coverage selection and policy structure, not about gathering the basics. The agent receives a structured submission with the class code, indicative premium, and any flagged endorsement needs (cyber, EPLI, professional liability) already surfaced. This same hand-off architecture is what makes AI for insurance agencies in 2026 work from lead capture to renewals.

A useful analog: Hartford's intake is doing for commercial what GEICO did for personal auto in the late 1990s — collapsing the data-gathering phase from a 20-minute phone interview to a self-serve flow, then routing the qualified prospect to the channel best positioned to close. The difference is that GEICO went direct-to-consumer, while Hartford is using AI to make its agent channel more productive, not to disintermediate it. The full story of how a similar direct-conversational shift has played out in auto is documented in GEICO's AI chatbot strategy for replacing forms in 2026.

Bundling BOP, workers' comp, and cyber via conversation#

Conversation lets Hartford bundle BOP, workers' comp, and cyber into a single intake without ever asking the owner to navigate three separate applications. Bundling is the single most important small commercial economic lever — retention for bundled small-business policyholders runs 15–20 percentage points above mono-line, and average premium per account roughly doubles when workers' comp is attached to a BOP. The historical problem has been that bundling required three separate applications, three classification flows, and three sets of underwriting questions. Owners gave up.

Hartford's conversational intake reuses 70–80% of the information across all three products. Once the AI knows the business is a 5-employee plumbing contractor with $400K in payroll and one company vehicle, the marginal work to extend the conversation from BOP into workers' comp ("Are any of your employees doing roof work above 15 feet?") and cyber ("Do you store customer credit card numbers or accept ACH payments?") is small — a handful of targeted questions, not a new 40-field form. This same multi-product conversational intake pattern is what makes Nationwide's AI customer experience for bundled insurance go conversational work — and what gives Branch Insurance's AI-native member experience for bundled policies its retention edge.

Three concrete bundling lifts conversational intake unlocks:

1. Workers' comp attach. Hartford reports workers' comp is its largest small commercial line. Conversational intake raises attach rates because the AI asks the W/C-triggering questions inside the BOP flow rather than forcing a separate application.
2. Cyber attach. Cyber attach on BOP runs well below 30% industrywide. Conversational intake surfaces the question naturally — "you mentioned you take credit cards from customers; do you want to add cyber coverage for breach response?" — and lifts attach materially.
3. EPLI for businesses with employees. The moment the owner mentions employees, the AI can offer Employment Practices Liability without the owner ever having heard the acronym.

This is the same pattern playing out in life insurance, documented in how conversational underwriting is replacing 90-page life applications and in workers' comp specifically, where Pie Insurance's AI-first workers' comp underwriting has proven the model scales.

What this signals for the $130B U.S. small commercial market#

The Hartford's bet signals that the $130 billion U.S. small commercial market will be re-segmented over the next five years by which carriers can run conversational intake at scale. McKinsey's 2024 small commercial insurance report pegs the U.S. small commercial premium pool at roughly $130 billion, growing 4–6% annually. The same report flags that only 20–25% of small commercial premium currently flows through digital-first underwriting — meaning 75–80% of the market is still served through traditional, form-based, agent-mediated workflows.

That gap is the prize. Three implications worth tracking:

• InsurTechs will not win the entire market. Next, Pie, Coalition, Hiscox, and Embroker have proven the model works — but the carriers with existing distribution, claims operations, and balance sheet (Hartford, Travelers, Chubb, Liberty Mutual, Nationwide) are now fielding the same conversational intake without giving up the moat. See Travelers Insurance's AI risk modeling and conversational underwriting shift and Chubb's AI strategy as a $260B specialty insurance leader for the parallel moves.
• Agents become higher-leverage, not obsolete. The carriers winning small commercial in 2030 will be the ones whose agents close 2–3x more policies per week because the intake is pre-qualified. This is exactly the same dynamic playing out in AI for insurance agents in 2026 with 64% adoption industry data.
• The class-code coverage race matters. Hartford's 1.3M-class appetite is a moat — but only if AI can actually classify into all 1.3M reliably. Carriers with narrower appetites can win specific lanes, but they cannot serve the long tail of unusual small businesses (food trucks, axe-throwing venues, mobile pet groomers) without conversational classification.

The bigger pattern — conversation replacing forms across every regulated, classification-heavy intake — is also playing out in healthcare, legal, and mortgage. See Rocket Mortgage's AI strategy for borrower intake for the closest analog outside of insurance, and the broader case for intelligent intake as a product surface for how this generalizes.

For carriers and brokers building this internally, the practical starting point is a conversational research project against your own SMB applicants. Built for product teams running a customer-discovery sprint, or built for CX teams running a voice-of-policyholder program, the insurance quote conversational template and the lead capture conversational template are starting points that work today.

Frequently Asked Questions#

What is The Hartford's small business insurance market share?#

The Hartford holds roughly an 11% share of the U.S. small commercial insurance market, making it one of the top 3 carriers in the small commercial segment by premium. The company reports approximately $5 billion in annual small business premium across more than 1.5 million policyholders. Its largest competitors in the segment are Travelers, Liberty Mutual, Nationwide, Chubb, and the rapidly growing AI-native carriers like Next Insurance and Pie Insurance.

What is a Business Owner Policy (BOP) and why is it hard to apply for?#

A Business Owner Policy (BOP) is a bundled insurance product combining general liability, commercial property, and business interruption coverage into one policy for small and mid-sized businesses. BOPs are hard to apply for because they require self-classification into NAICS or ISO class codes, payroll and revenue disclosures, property specifications, and prior loss information — typically 30–60 questions. Most owners cannot complete a BOP application unassisted, which is why conversational AI intake matters.

How does conversational AI improve commercial insurance applications?#

Conversational AI improves commercial insurance applications by letting the business owner describe their operation in plain language rather than self-classifying into industry codes. The AI extracts the relevant entities — class code, payroll, employee count, revenue, property details — and asks targeted follow-up questions only where it has low confidence. Industry data suggests conversational intake cuts application time from 22–35 minutes to under 7 minutes and lifts submission completion rates by 40–60% versus static forms.

Will AI intake replace insurance agents and brokers?#

AI intake will not replace insurance agents and brokers in small commercial — it will make them dramatically more productive. The Hartford, Travelers, and Nationwide all route conversational intake submissions to their independent agent channels with pre-qualified risk classification, appetite confirmation, and indicative pricing already attached. Agents spend their time on coverage selection, endorsement structure, and relationship building rather than data gathering. The net effect is 2–3x more bound policies per agent per week.

How is The Hartford different from Next Insurance or Pie Insurance?#

The Hartford differs from Next Insurance and Pie Insurance in two main ways: it operates as a multiline carrier with $26 billion+ in annual revenue across personal, small commercial, mid-market, group benefits, and specialty lines, and it sells primarily through independent agents rather than direct-to-consumer. Next and Pie are AI-first InsurTechs focused on small commercial direct sales. The Hartford's competitive response has been to fold AI-native conversational intake into its existing agent channel rather than launch a separate direct brand.

What industries does The Hartford's BOP cover?#

The Hartford's Business Owner Policy covers more than 1.3 million classes of business, making it one of the broadest small commercial appetites in the U.S. market. Covered industries include retail, restaurants, professional services, contractors and trades, light manufacturing, wholesalers, healthcare practices, technology services, and specialty operations. The breadth of class-code coverage is one of the main reasons conversational classification matters — narrow-appetite carriers can use simpler intake, but covering 1.3M classes requires AI that can map natural-language business descriptions to the right code.

Conclusion: Commercial insurance AI is a small-business unlock, not an enterprise toy#

The Hartford's small commercial AI investment is the clearest signal in the U.S. market that commercial insurance AI is now a small-business unlock, not an enterprise pilot. A top-10 carrier with 11% market share, $5B in small business premium, and a 215-year-old brand does not bet the small commercial book on conversational intake unless the underlying economics are real. They are. Conversation collapses the application from 30 minutes to 5, lifts BOP-to-workers'-comp attach, surfaces cyber and EPLI for businesses that would never have asked for them, and routes a pre-qualified submission to the agent or broker channel that closes the policy.

For other carriers, MGAs, and digital agencies, the practical playbook is: pick the line where application friction is highest, replace the form with a conversation, route the structured output to your existing channel, and measure the lift in submission-to-bound conversion. That is what The Hartford is doing. That is what AIG, Travelers, Chubb, and Nationwide are doing in parallel. The carriers that don't move will keep losing the long tail of unusual small businesses to InsurTechs that solved the intake problem first.

If you're at a carrier, broker, MGA, or InsurTech building this internally, run a conversational research project against your own SMB applicants — you'll learn more about why your applications get abandoned in two weeks than your CX team has learned in two years. See pricing or browse studies from carriers, brokers, and agencies already using Perspective AI to replace static intake with conversation.