Thursday, November 12, 2026

Today’s Edition

AI Intel Report

MARKETS —

AI Agents

HDI Seguros Cuts Insurance Product Development Time 85% With AI Agents

Brazilian insurer and insureMO apply human-in-the-loop generative AI agents to life and affinity product engineering after a record Sompo portfolio migration, launching 19 products and passing R$500 million in premiums.

10 MIN READ
Team in HDI Seguros office reviewing policy documents while a manager gestures near a window.
Illustration: AI Intel Report

Human-in-the-loop generative AI agents are software systems that translate natural-language product requirements into structured specifications, application programming interfaces, quotation interfaces, and end-to-end test workflows, with explicit human approval required before any write operation or execution, the engineering model HDI Seguros and insureMO are applying to life and affinity insurance products.

Executive Summary

HDI Seguros, one of Brazil's largest insurers, and insureMO, a global insurance middle office platform provider, are deploying generative AI agents across the insurance product engineering lifecycle. The companies announced the next phase of their strategic partnership on Nov. 4, 2026, and the early result is a roughly 85% reduction in development and configuration time, according to PR Newswire. The project has launched 19 commercial life and affinity insurance products, and cumulative premiums on the new life insurance core system have exceeded R$500 million, approximately US$93.5 million.

The AI deployment sits on a migration that was itself a benchmark: HDI Seguros integrated the life insurance portfolio acquired from Sompo onto insureMO's architecture in less than nine months, moving more than 11 million records spanning policies, endorsements, claims, and invoices. That system has served as HDI Seguros' life insurance core since 2023. The generative agents now operate on top of that production platform, compressing the time between product conception and launch.

For C-suite readers, the case translates agentic AI into a concrete enterprise win: measurable time savings, a growing premium base, and a governance model that keeps humans in control. The insurer's product pipeline is the business outcome; the AI agents are the mechanism.

Why did HDI Seguros and insureMO target product engineering with AI agents?

Insurance product engineering is one of the most configuration-intensive processes in financial services. A new life or affinity product requires structured specifications, application programming interfaces, quotation and policy-issuance screens, end-to-end workflow testing, and documentation, each step traditionally handled by specialists working across core systems. In a middle office platform, that routine configuration work consumes weeks per product.

InsureMO operates as an insurance middle office platform, the layer between distribution channels and core insurance systems where products are assembled, configured, and managed. The partnership's stated objective, per PR Newswire, is to bring generative AI agents to the engineering and product factory lifecycle for insurance products. Revista Apólice, the Brazilian insurance trade publication, also reported the new phase of the partnership, describing the application of generative AI agents to product development and configuration. The product factory concept is deliberate: it treats product creation as a repeatable pipeline rather than a bespoke project.

The Brazilian market context makes speed strategically valuable. Life and affinity insurance lines in Brazil are distributed heavily through partners such as banks, retailers, and brokers, where carriers compete on how quickly they can launch and adapt products. An 85% reduction in engineering time directly changes that competitive calculus.

What did the Sompo portfolio migration establish?

The migration is the technical foundation for the AI deployment. HDI Seguros brought the life insurance portfolio acquired from Sompo onto insureMO's architecture in less than nine months, according to Insurtech Insights. The effort moved more than 11 million records, including policies, endorsements, claims, and invoices, and the platform has operated as the insurer's life insurance core system since 2023.

The significance is that this was a production migration at operating scale, not a greenfield implementation. Vanesa Bustamante, vice president of technology at HDI Group, described the migration as proof that a large insurer can modernize its architecture while remaining live. The resulting platform gave HDI Seguros the scale and flexibility needed to bring new products to market faster, she said.

The migration also produced the data foundation the AI agents depend on. Product engineering on a modern platform can leverage structured product models, existing API patterns, and test frameworks, the raw material generative agents need to produce specifications and interfaces that match the insurer's operating standards.

What tasks do the generative AI agents perform in the engineering lifecycle?

The agents operate as a pipeline from natural-language requirements to deployable product artifacts. Rafael Rodrigues, general manager for Latin America at insureMO, described the workflow: agents structure the specifications, generate the necessary APIs, create quotation and policy-issuance interfaces, perform end-to-end workflow testing, and document the process.

  1. Structure product specifications from requirements described in natural language.
  2. Generate the application programming interfaces required by the new product.
  3. Create quotation and policy-issuance interfaces for distribution partners and customers.
  4. Perform end-to-end workflow testing across the product lifecycle.
  5. Document the process for audit, traceability, and operational continuity.

The mechanism matters as much as the task list. A product specialist describes requirements in natural language, and the agents convert that description into engineering artifacts that previously required manual configuration. Because the outputs follow the platform's established patterns, the generated specifications and interfaces are consistent with the insurer's existing product portfolio.

Rodrigues is explicit about what the agents do not do. AI agents do not replace product specialists, he said. They eliminate weeks of routine configuration work, freeing specialists to focus on the areas that require judgment: defining risk and pricing. That division of labor is the design principle: machines handle the deterministic configuration, humans retain the judgment-intensive decisions.

How is governance structured for the human-in-the-loop model?

The governance architecture is what makes the speed safe. InsureMO's deployment defaults to simulation: proposed changes run in a simulated environment before any production effect. Write operations and execution require explicit human approval, and all operations are reversible, allowing failed changes to be rolled back without production impact.

That combination addresses the failure modes that typically slow enterprise adoption of agentic AI: hallucinated configurations, unauthorized writes to production systems, and unrecoverable errors. By making autonomy conditional on reversibility, the model fits inside existing change-management and audit frameworks rather than requiring new ones.

Bustamante tied the control layer directly to the business case. The company is dramatically reducing the time between product conception and launch while maintaining full security and traceability, she said. Traceability, in a regulated sector, is not a compliance nicety; it is a precondition for allowing AI-generated artifacts into the product lifecycle.

What measurable results has the deployment produced?

The headline metric is the approximately 85% reduction in development and configuration time, reported by PR Newswire. Applied across 19 launched products, the velocity gain compounds: engineering teams can evaluate more product variations, respond faster to distribution partners, and retire the backlog of product ideas that manual configuration made uneconomical.

Insurance product engineering before and after generative AI agents
StageTraditional approachWith AI agents
SpecificationSpecialists manually draft structured product specificationsAgents structure specifications from natural-language requirements
API generationEngineers build and configure APIs by handAgents generate the required APIs automatically
Interface creationQuotation and policy-issuance screens configured manuallyAgents create quotation and policy-issuance interfaces
Workflow testingManual end-to-end testing across systemsAgents perform end-to-end workflow testing
DocumentationWritten manually, often at project closeAgents document the process continuously
Time to launchWeeks of routine configuration per productApproximately 85% reduction in development and configuration time

The commercial results give the efficiency gain a revenue dimension. Cumulative premiums on the new life insurance core system have exceeded R$500 million, approximately US$93.5 million, with monthly premiums of roughly R$30 million, per PR Newswire. The companies project premium volume to reach R$1 billion in 2027. Those figures indicate that a faster product pipeline is converting into written business, not just engineering throughput.

The metric mix is worth noting for executives: the deployment is measured in engineering time, product count, and premium volume simultaneously. That triple measurement, efficiency, output, and revenue, is the difference between an AI pilot and an AI business case.

What do the executives say about the AI-agent rollout?

Migrating the life insurance portfolio and integrating new operations demonstrate that it is possible to modernize a large insurer's architecture while it remains in production, without disrupting day-to-day service to our partners and customers. Our partnership with insureMO has given us the scale and flexibility needed to bring new products to market faster. Now, by integrating AI agents into the engineering lifecycle with strict governance and human oversight, we are dramatically reducing the time between product conception and launch while maintaining full security and traceability.Vanesa Bustamante, Vice President of Technology at HDI Group

Bustamante's statement connects three phases of the engagement: the migration proved modernization is possible in production; the platform partnership delivered scale and flexibility; and the AI layer accelerates the product pipeline under governance. The sequence is a template for insurers that have postponed core modernization out of operational risk concerns.

Based on requirements described in natural language, AI agents structure the specifications, generate the necessary APIs, create quotation and policy issuance interfaces, perform end-to-end workflow testing, and document the process. AI agents do not replace product specialists. They eliminate weeks of routine configuration work, freeing specialists to focus on the areas that require judgment: defining risk and pricing.Rafael Rodrigues, General Manager, Latin America at insureMO

Rodrigues' framing addresses the workforce question that dominates enterprise AI planning. The agents absorb routine configuration; specialists keep the judgment-intensive work of risk and pricing. For a C-suite reader, that is the difference between labor substitution and labor leverage, and labor leverage is the model that tends to survive contact with regulated production environments.

How does this deployment compare with other enterprise AI-agent initiatives?

Most documented agentic AI deployments in large enterprises to date have targeted internal functions such as customer support, software code generation, and document processing, where the cost of an error is contained. The HDI Seguros case is distinct because the agents produce artifacts in a regulated product lifecycle with direct revenue consequences. The 85% time reduction is tied to a premium base measured in hundreds of millions of reais.

The governance design is also ahead of the typical pilot. Many enterprises test AI agents in read-only or advisory roles before granting write access; insureMO's model formalizes that staging as a permanent control, simulation by default, human approval for writes, and reversible operations, rather than a temporary sandbox. That makes the control architecture a product feature rather than a pilot concession.

For peer insurers, the comparison point is sobering: competitors with modern middle office platforms can now overlay generative agents on product engineering, while carriers still running product configuration on legacy cores face a widening velocity gap. The strategic window for platform modernization is defined by exactly this kind of capability differential.

What are the implications for insurers and enterprise technology leaders?

For insurers, the implication is competitive. Product time-to-market is a strategic variable in life and affinity lines, where distribution partners select carriers on speed and flexibility. An 85% reduction in engineering time changes the economics of testing many small products, which matters in affinity distribution where partner-specific products are the norm.

For chief information officers, the transferable asset is the sequencing: modern platform first, agentic AI second. The generative agents succeeded because the insureMO architecture gave them structured product models, API patterns, and test frameworks to work with. Deploying agents on an undocumented legacy core would not have produced the same result.

For chief executive officers, the case reframes agentic AI from an abstract capability to a measured business outcome. The project pairs a headline efficiency metric, 85%, with revenue evidence, R$500 million in cumulative premiums, and a risk framework that keeps humans accountable. That combination is what boards ask for before approving AI programs at scale.

What comes next for HDI Seguros and insureMO?

The companies describe the AI integration as the next phase of the partnership, not an endpoint. With premium volume projected to reach R$1 billion in 2027, the product factory will need to sustain a higher cadence of launches, and the AI agents are the capacity lever for that pipeline. The R$30 million monthly premium run rate is the near-term operating signal to watch.

The trajectory also signals where the insurance middle office category is heading: platforms that can ingest a legacy portfolio, run it in production, and then overlay generative agents on the engineering lifecycle. InsureMO's expansion in Latin America, anchored by one of Brazil's largest insurers, gives the reference case weight in a market where insurance distribution is digitizing rapidly.

The open questions are durability and scale. Whether the 85% time reduction holds as products grow more complex, and whether the governance model extends as agent autonomy increases, will determine how far the approach spreads. For peer executives, the immediate lesson is the sequence: production modernization, platform consolidation, then agentic AI under human-in-the-loop control.

Frequently asked questions

Frequently asked

What are generative AI agents in insurance product engineering?

Generative AI agents are software systems that translate natural-language product requirements into structured specifications, APIs, quotation and policy-issuance interfaces, end-to-end workflow tests, and documentation. In the HDI Seguros deployment, they operate under a human-in-the-loop model in which write operations and execution require explicit human approval.

How much time did HDI Seguros save using the AI agents?

HDI Seguros and insureMO report an approximately 85% reduction in development and configuration time for insurance products, according to PR Newswire. The project has launched 19 commercial life and affinity insurance products on the new core system.

What was the Sompo portfolio migration, and why does it matter?

HDI Seguros integrated the life insurance portfolio acquired from Sompo onto insureMO's architecture in less than nine months, moving more than 11 million records including policies, endorsements, claims, and invoices. The migration established the production platform on which the AI agents now operate, serving as HDI Seguros' life insurance core system since 2023.

How does the human-in-the-loop governance model work?

AI agents run proposed changes in simulation by default. Write operations and execution require explicit human approval, and every operation is reversible, allowing failed changes to be rolled back without production impact. The model preserves security and traceability while automating routine configuration work.

Sources

  1. PR Newswire / INSUREMO — Reports approximately 85% reduction in development and configuration time, cumulative premiums exceeding R$500 million (about US$93.5 million), monthly premiums near R$30 million, and projection of R$1 billion in premium volume by 2027; announcement of the next phase of the partnership on Nov. 4, 2026.
  2. Insurtech Insights — Reports the launch of 19 commercial life and affinity insurance products and the migration of more than 11 million records in less than nine months, describing the AI deployment as the next phase of the companies' strategic partnership.
  3. Revista Apólice — Reports the new phase of the HDI Seguros-insureMO technology partnership applying generative AI agents to the development and configuration of insurance products, including the up-to-85% time reduction.