Agentic AI is coming - should brokers be worried?
Short answer: no-if you adapt. At ITC London 2026, senior leaders across insurance and tech agreed that AI agents will rework workflows, shift where value sits, and force operating model change. Human expertise remains the differentiator. The firms that win will pair autonomy with accountability.
From automation to autonomy: the 80/20 reality
Most carriers and brokers are still moving from manual to automated and augmented processes. The next phase is autonomy-systems that take action, not just suggest it.
McKinsey estimates automation and AI-assisted workflows can deliver 20-30% productivity gains in underwriting and claims by removing manual data handling and rekeying. Fully autonomous decisions, especially in complex commercial risks, remain out of reach-nuance and context still carry the day.
Think "80/20." AI agents excel at repetitive, data-heavy steps-extracting submission data, enriching records, pre-populating files. The tough final 20% needs judgment, negotiation, and context that lives in people, not PDFs.
What changes for brokers and carriers
Broker placement is a prime target. AI can rank markets by likelihood to quote, flag missing data, and automate clarification loops. That lets brokers approach more carriers with the same effort and cut cycle time.
On the carrier side, large language models are accelerating long-discussed goals: machine-readable policies and computable contracts. That creates tighter links across pricing, claims, and reserving-and compresses handoffs that slow responsiveness today.
Reality check: limits still matter
Today's large language models do not fully handle multi-sensory inputs or higher-order abstractions. Technical hurdles are real, and some may take years to crack. Treat autonomy as a staged rollout with guardrails, not a switch you flip.
Culture and talent will make or break it
With a significant share of the workforce nearing retirement, capturing institutional knowledge is urgent. The message from leaders: position AI as a teammate, not a replacement. Resistance to change and data issues-not algorithms-are often the main blockers.
"We can have phenomenal technology in this industry, but culture and talent are what determine whether that technology actually gets used, and used in the right way," said Scott Sayce, chief innovation officer at DUAL Group. Move fast, but not so fast that you lose your best people.
A two-phase playbook (learn from Rolls-Royce)
Phase one is incremental: use sensors (or in our case, AI) to cut costs and improve reliability. Phase two rethinks the model itself. Rolls-Royce didn't stop at cheaper maintenance-they shifted to engines-as-a-service, priced by usage.
Expect a similar arc in insurance. First, speed up intake and triage. Then, rethink distribution, pricing triggers, and service models once the constraint (manual work) is removed.
How "Power by the Hour" changed the engine business
Practical next steps (90-day plan)
- Map three workflows with clear friction: broker placement, submission triage, and claims FNOL to adjudication.
- Deploy narrow AI agents where rules are clear: document extraction, data enrichment, quality checks, eligibility triage.
- Set decision guardrails: thresholds for auto-approve/auto-decline, human-in-the-loop criteria, and audit trails.
- Fix data at the source: standardize intake, define golden fields, and create feedback loops to reduce missing info.
- Pilot one broker placement agent with 5-10 markets; measure quote rate, cycle time, and hit ratio.
- Stand up a change squad: a broker lead, an underwriter, a claims lead, compliance, and one engineer who ships.
- Upskill your team: prompts, QA of AI output, exception handling, and client communication on how AI is used.
What this means for brokers
- Your edge shifts to judgment, relationships, and exception handling. Let AI handle the drudge work.
- Use market-ranking agents to expand carrier reach without adding email ping-pong.
- Lean into data: cleaner submissions and clearer narratives raise your close rate more than a bigger spreadsheet.
- Track metrics that matter: time-to-first-quote, completion rate of data requests, bind ratio by market segment.
Governance that keeps regulators comfortable
- Document model purpose, data sources, and known limits.
- Keep humans in the loop for pricing, coverage determinations, and declinations.
- Log all automated decisions and surface explanations that a client (and auditor) can understand.
Where to build skills
If your team needs a quick ramp on AI skills by role, explore focused learning paths and certifications that map to day-to-day work.
Bottom line
Agentic AI will rework insurance workflows and move value to those who execute with discipline. The winners won't be the first to demo a bot-they'll be the ones who pair automation with clear guardrails, better data, and people who know when to override the system. Brokers who embrace that mix won't be replaced; they'll be in higher demand.
Your membership also unlocks:
