AI to E&S: Getting Ahead of the Hard Market, Closing Protection Gaps, and Rethinking Cyber Risk

Reshaping distribution and underwriting with AI

Distribution and underwriting are moving from batch work to real-time decisions. The winners will collapse intake friction, enrich data at the point of quote, and score submissions against appetite instantly.

• Unify data: broker emails, ACORDs, loss runs, valuations, third-party hazard data. Make it searchable and standardized.
• Score every submission: propensity-to-bind, loss probability, fit to appetite, premium adequacy. Route high-value risks to humans.
• Stand up appetite APIs and digital guidelines so partners know what you want-and what you don't-before they submit.
• Prefill apps with public and commercial data to reduce questions and speed quotes.
• Use human-in-the-loop review for exceptions and emerging risks.

How to get ahead of the hard market

Stop reacting at renewal. Bring structure early, show credible alternatives, and prove loss cost control. Hard markets reward preparation and clarity.

• Market 90-120 days out with a one-page risk thesis: exposures, controls, loss story, and requested structure.
• Resegment accounts: layer, increase retentions, and target markets by risk factor-don't spray and pray.
• Add engineering evidence: photos, sensor data, water/leak logs, cyber control attestations.
• Consider parametric cover for named perils and use captives/fronting for frequency layers.
• Prepare board-ready options: status quo, structure change, and E&S path with trade-offs.

How brokers can explain higher prices

Clients accept price when they see the inputs. Be transparent on loss trends, reinsurance costs, exposure growth, and model changes.

• Break down TIV, location count, payroll, fleet miles, or revenue changes line by line.
• Share loss triangles and benchmark severity/frequency versus peers.
• Quantify new costs: social inflation, CAT load, reinsurance rate-on-line, and modeling updates.
• Tie controls to dollars: "MFA reduced ransomware expected loss by $X; sensors cut water losses by $Y."

How distribution and underwriting will change

Underwriting shifts from document review to portfolio steering. Distribution moves where the customer is-embedded, partner-led, and API first.

• Continuous underwriting: ingest new signals (valuations, sensors, credit, mobility) and adjust pricing midterm.
• Underwriters focus on exceptions, referrals, and accumulation management-not data entry.
• Small commercial and personal lines skew to digital; complex risk remains advisory but with better prefill and triage.
• Distribution broadens: MGAs with specialized data, retailers with embedded offers, and carrier direct for simple risks.

Practical AI uses in insurance today

AI is most useful where there's repetitive text, unstructured intake, or pattern detection. Start where the data and payback are clear.

• Submission ingestion: read ACORDs, SOVs, loss runs; extract and normalize fields.
• Entity resolution: dedupe names/addresses; flag related entities and prior loss history.
• Risk enrichment: geocode locations; add hazard, crime, and fire protection data; run cyber surface scans.
• Claims: triage severity, automate FNOL, detect fraud patterns, and recommend reserves.
• Governance: document models, monitor drift, test bias, and keep a human in final decisions.

For model governance and risk controls, see the NIST AI Risk Management Framework.

Proving value in cyber beyond price

Cyber buyers want fewer incidents and faster recovery. Price is one lever. Controls and readiness are bigger ones.

• Baseline against the NIST Cybersecurity Framework: identify, protect, detect, respond, recover.
• Deliverables that land: external surface scan, MFA/EDR status, backup health, and user training rates.
• Quantify impact: expected annual loss, top three threat scenarios, downtime days, and control ROI.
• Run a tabletop and document lessons; tie corrective actions to coverage terms and pricing credits.

Delving into the future of insurance - from AI to E&S

E&S keeps growing because it solves for speed, specialization, and capacity constraints. AI will make that specialization even sharper.

• MGAs with proprietary data win niche classes and speed-to-quote.
• Carriers lean on E&S to flex appetite, manage accumulation, and test new products.
• Expect hybrid journeys: admitted for core risks, E&S for gaps or unique exposures.

How carriers and brokers can work together better

Alignment beats volume. Share data, clarify appetite, and give feedback loops that improve quotes and outcomes.

• Publish appetite in plain language with yes/no examples and circulate weekly updates.
• Quarterly file reviews: declination reasons, hit/miss analysis, and bind blockers.
• Joint loss control: pre-bind walkthroughs, post-loss reviews, and measurable corrective actions.
• Standardize SOV formats, valuations, and CAT coding to cut rework.

Protection gaps, catastrophe losses, and the E&S market

Protection gaps widen with inflation, valuation errors, and secondary perils. E&S fills some needs but isn't a cure-all.

• Fix the basics: current valuations, precise geocoding, defensible BI calculations.
• Mitigation matters: roof upgrades, flood barriers, defensible space, and water sensors reduce loss cost and improve options.
• Use parametric add-ons for quake, wind, or flood to protect cash flow and deductibles.
• Be clear on CAT aggregates and client accumulation exposure; avoid surprise non-renewals.

What we've learned from recent CAT losses

Models missed frequency and severity of secondary perils; supply chain inflation amplified severity; and undervaluation hurt recoveries. Data quality and mitigation beat wishful thinking.

• Update valuations annually; document the method and sources.
• Invest in sensors and maintenance logs to show reduced hazard and faster detection.
• Stress-test coverage: sublimits, waiting periods, ingress/egress, and service interruption.
• Pre-arrange vendors and debris removal; speed matters to loss size.

Could AI reduce losses?

Yes, where you can detect early signals or change behavior. The key is instrumenting exposures and turning alerts into action.

• Property: leak, temperature, and vibration sensors with clear escalation paths.
• Auto: telematics coaching, distracted driving alerts, and repair network optimization.
• Workers' comp: early triage, nurse hotlines, and fraud red flags.
• Tie verified controls to pricing credits and share savings math with clients.

Should government step in on cyber risk?

Systemic cyber events can exceed private capacity. A public backstop, clear data-sharing rules, and safe harbors for best practice could stabilize pricing and capacity.

• Backstop structure similar to terrorism cover to handle correlated, nation-state-scale events.
• Mandatory incident reporting and anonymized data sharing to improve modeling.
• Safe harbor for firms meeting baseline controls to encourage adoption.

Building a people-first culture in insurance

AI and automation reduce busywork, but trust moves business. Give teams better tools, protect focus time, and celebrate decisions that improve client outcomes-quota follows.

• Train producers and underwriters on data tools and prompt skills, not just platforms.
• Rewrite incentives around quality: retention, loss ratio, and control adoption-not submissions dumped.
• Form small squads (broker, underwriter, claims, engineer) around key accounts with shared goals.

Carrier-broker execution checklist

• Data standard agreed (SOV, CAT codes, valuations, loss runs).
• Appetite clarity with live updates and example accounts.
• Submission SLA: time to quote, bind, and reasons for decline.
• Loss control feedback loop with measurable fixes and timelines.
• Claims review cadence with trends and reserve transparency.
• AI use policy: data rights, model monitoring, and human oversight.

Metrics that matter

• Quote speed, hit ratio, retention, and average days to bind.
• Loss ratio by segment and by control adoption (with/without sensors, MFA, etc.).
• Claim cycle time and indemnity severity versus benchmarks.
• CAT-exposed TIV to surplus and reinsurance spend to GWP.
• Cyber control pass rate and incident frequency per 1,000 employees.

Where to skill up on AI

Give your team practical reps with tools they'll use on submissions, claims notes, and client reports. Start small, ship fast, measure impact, then scale.

• Latest AI courses for quick wins your team can apply to intake, analysis, and reporting.
• Courses by job to upskill producers, underwriters, and claims pros without fluff.

Insurance rewards clarity. Use data to prove risk quality, use AI to cut friction, and use people to build trust. Do those three and you won't just get through a hard market-you'll grow in it.