What dynamic insurance pricing with an AI layer means
Dynamic insurance pricing with AI is the practice of calculating the premium (prêmio) in real time from a multi-factor risk score for the individual risk, rather than reading it off a fixed rate table. The premium reflects the specific exposure of each case and updates as data, loss experience, and risk appetite (apetite de risco) change. For Brazilian Seguros e Danos (P&C) insurers, this matters most in commercial and specialty lines (ramos) such as property, engineering, transport, and liability, where each risk is assessed individually and a single static table cannot capture the spread of exposures.
The buyer for this is an underwriting (subscrição) lead, a product head, or an innovation team deciding whether to automate the pricing stage of the quotation and underwriting journey. The shift they are weighing is from periodic table updates, applied by hand and prone to underwriter-to-underwriter variance, toward a priced quote that comes back firm and risk-adjusted in the moment the broker (corretor) submits it.
An external AI layer makes this possible without touching the actuarial core. It sits on top of the insurer's existing core system and actuarial engine, connects by API, quotation portal, or document upload, structures the incoming submission, scores the risk, applies the insurer's own pricing logic to that score, and writes the decision back with a full audit trail. The actuarial team still owns the rating basis and the technical premium. The layer automates how that policy is applied to each quote, calibrated to the insurer's underwriting manual (manual de subscrição). This is the WIR posture, the AI layer for insurance, on top of the systems the insurer already runs, never in their place. It does not carry risk and it is not an actuary.
How the risk-adjusted premium is calculated in real time
Dynamic pricing is one stage in a connected sequence, fed by the stages before it and feeding the decision after it. The journey runs in six stages. First, multichannel intake normalizes every submission that arrives by API, broker portal, email, or uploaded document into a single structured record, so nothing has to be re-keyed. Second, intelligent document reading uses Machine Learning and document AI to extract the insured data, the object at risk, location, occupancy, values at risk (importância segurada), and prior loss history from unstructured PDFs and spreadsheets. Third, broker enrichment and scoring adds external and internal context such as location exposure, sector, prior claims, and compliance checks, and produces a completeness score so incomplete submissions are flagged or auto-completed.
The fourth stage is the risk and fraud Machine Learning engine, which produces a multi-factor risk score for the individual risk by combining the structured submission, the enrichment data, and the insurer's historical loss patterns, scoring fraud and anomaly signals in the same pass. The fifth stage is dynamic pricing itself, and it is the focus of this guide.
Here the premium is derived from the risk score, not read off a fixed table. The technical premium, the prêmio puro or technical rate, comes from the actuarial core or the rating tables the actuarial team owns. The AI layer applies the multi-factor adjustment to that technical base in real time, per the insurer's underwriting manual. Pricing is multi-factor, because dozens of correlated variables such as object, location, occupancy, values at risk, deductible (franquia), coverage scope, loss history, and sector influence the premium at once, which a static table cannot do. Pricing is calibrated to risk appetite, so risks outside appetite are loaded, referred, or declined rather than priced into the book silently. Pricing is real time, so the broker receives a firm, risk-adjusted number in the moment instead of waiting for manual rating. Pricing is consistent, because the same risk profile yields the same premium every time, removing underwriter-to-underwriter variance. The layer can express the premium with the commercial levers the insurer allows, such as commission, discount bands, and loadings, all bounded by the insurer's own policy so nothing is priced outside the manual.
The sixth stage is decision and prioritization with an audit trail. The layer routes the quote straight-through for clean, in-appetite, in-authority risks, and refers cases that need human judgment to an underwriter with the score, the priced premium, and the contributing factors attached. Every quote carries a full audit trail of the data used, the model version, the factors that drove the score and the price, and the decision path. The result is a premium that reflects the true exposure of the individual risk, calculated automatically, while the actuarial rating basis and the risk itself remain entirely with the insurer.
How to deploy dynamic pricing as an external layer
Deploying dynamic pricing as an external AI layer is additive and reversible. It consumes data the insurer already produces and returns structured output the insurer's systems already understand, so there is no core migration and no IT project for the insurer's team to run. The contrast is with swapping a core insurance system, a multi-year, high-risk program that IT and legacy-system limitations frequently block. According to BCG, 70% of insurers do not execute innovation precisely because of IT limitations, which is the reason an external layer matters here.
A practical sequence runs in six steps. The first is scope and line selection: pick a starting ramo where pricing pain is highest and volume justifies automation, often a high-frequency line such as auto or a high-friction specialty line such as property, then define the journey to automate and the SLA target. The second is integration with the core, connecting the layer to the core and the broker channel by API, portal, or upload, where the core stays the system of record, data flows in, and structured priced quotes and decisions flow back. The third is calibration to the underwriting manual and risk appetite, encoding the insurer's underwriting rules, appetite boundaries, authority limits, referral triggers, and pricing levers. The technical rating basis comes from the actuarial team, and the layer is calibrated to apply it, which is what makes the priced quote the insurer's own policy expressed automatically rather than a black box.
The fourth step is testing and back-testing, running the model against historical submissions and known outcomes, comparing priced premiums against the existing table and against realized losses, and tuning until pricing is consistent, in-appetite, and explainable. The fifth is a controlled go-live, starting in a shadow or assisted mode where the layer prices and the underwriter confirms, then expanding straight-through authority as confidence builds while keeping human referral for out-of-appetite and high-value risks. The sixth is continuous operation and monitoring of pricing accuracy, loss ratio by segment, conversion and SLA, model drift, and exception rates, recalibrating as loss experience and appetite evolve. Setup runs 3 to 12 months as a fixed-scope implementation with KPIs agreed before start, after which the platform moves into continuous operation. Throughout, the actuarial core is never touched and the insurer runs no IT project of its own.
Governance, explainability, and LGPD
Automated pricing in Brazil must be explainable, auditable, and compliant, and those requirements shape how the layer is built rather than being added afterward. Every premium has to be traceable to the factors that produced it, which data, which risk-score drivers, which loading or discount, and which model version. A price an underwriter or auditor cannot reconstruct is not acceptable for a regulated technical premium, so the layer surfaces the contributing factors for each quote. Each quote also carries a full audit trail of inputs, enrichment sources, score, priced premium, decision path, timestamps, and the policy version applied, which supports internal audit, SUSEP supervision, and dispute resolution.
LGPD (Lei Geral de Proteção de Dados, Lei nº 13.709/2018) governs the personal data that pricing uses. It requires a lawful basis for processing, purpose limitation, and data-subject rights, and its Article 20 gives the data subject the right to request review of decisions taken solely on the basis of automated processing that affect their interests, which applies directly to automated underwriting and pricing. This is exactly why explainability and a human-referral path are designed in. The full text of the law is available from the official Planalto source, and the data-protection authority publishes guidance through the ANPD portal. SUSEP supervises conduct in the insurance market through its official channel.
Submission data, enrichment data, and priced quotes are encrypted in transit and at rest, because the layer handles sensitive personal and risk data across the whole journey. Governance is ultimately anchored in one fact: the model applies the insurer's own underwriting manual and actuarial rating basis. The insurer remains accountable for the price and the risk, and the layer makes that policy faster and more consistent while keeping it explainable and auditable. Dynamic pricing does not replace the actuarial core. It automates the application of it.
How WIR automates dynamic pricing
WIR Innovation is the AI layer for insurance, an external AI platform that automates the quotation and underwriting journey on top of the insurer's existing systems in Brazil. It is 100% external, with no load on the insurer's IT and no core migration, and its Machine Learning is calibrated to each insurer's risk appetite and underwriting manual. WIR is not an insurer, a broker, or an MGA, and it does not carry risk. It automates the pricing step per the insurer's own risk-acceptance policy, and the actuarial team keeps the rating basis and the technical premium.
Two modules carry most of the pricing work. Underwriter Intelligence automates the quotation journey per the insurer's risk policy, with real-time ML scoring calibrated to appetite, automatic routing by appetite and exposure, and predictive conversion analysis by product, risk, and broker, so underwriters spend their time on risk analysis and business development rather than manual rating. Smart Sales is distribution intelligence that maps the portfolio by client and product, scores upsell and next-best-action, and runs multi-channel campaigns with an attribution trail, so penetration and retention grow together. Real-time dashboards and analytics give a proactive view of in-flight deals and the pipeline. In WIR's six-stage platform flow, dynamic pricing produces the risk-adjusted premium as an instant output, then the decision stage returns a quote, an automatic decline, or an escalation to a human, always with an explanation, writing back to the policy core with a full audit trail.
The competitive context is concrete. The Seguros e Danos market grows double digits per year, while company structure does not keep pace with that acceleration. Deloitte finds underwriters spend 40% of their time on administrative tasks, Capgemini reports that 60%+ of brokers choose an insurer by response speed, and Gartner estimates corporate teams lose 20-30% of their time organizing unstructured data. Real-time, risk-adjusted pricing addresses each of those pressures directly. WIR's only public traction at this stage is a first POC in execution with a global insurer in the Transport line. Every decision the platform makes is explainable and returns a full audit trail, and data is encrypted at every step and LGPD compliant. The AI layer for insurance. On top of the systems the insurer already runs, never in their place. To see where it fits, the team can map the pricing journey through a conversation with WIR.
Frequently asked questions
How is the premium calculated from the risk score?
The premium is derived from a multi-factor risk score, not read off a fixed table. WIR's risk and fraud Machine Learning engine scores the individual risk by combining the structured submission, enrichment data, and historical loss patterns. The AI layer then applies the insurer's own technical premium and pricing logic to that score in real time, calibrated to the underwriting manual. Dozens of correlated variables, such as object, location, occupancy, values at risk, and loss history, influence the premium at once.
Does dynamic pricing replace the insurer's actuarial core?
No. Dynamic pricing does not replace the actuarial core. It automates the application of it. WIR is an external AI layer on top of the insurer's existing systems, with no core migration and no IT project for the insurer to run. The actuarial team still owns the rating basis and the technical premium. The layer applies that policy to each quote in real time, calibrated to risk appetite. WIR is not an insurer, broker, or MGA, and it does not carry risk.
Is the AI-generated price explainable and auditable?
Yes. Every priced quote is explainable and carries a full audit trail. WIR surfaces the contributing factors behind each premium, which data, which risk-score drivers, which loading or discount, and which model version produced it. A price an underwriter or auditor cannot reconstruct is not acceptable for a regulated technical premium. The trail supports internal audit, SUSEP supervision, and dispute resolution, and it underpins the LGPD Article 20 right to request review of an automated decision. Data is encrypted at every step.
Is pricing kept calibrated to the insurer's risk appetite?
Yes. WIR's Machine Learning is calibrated to each insurer's risk appetite and underwriting manual. Risks outside appetite are loaded, referred, or declined rather than priced into the book silently. The Underwriter Intelligence module applies real-time scoring calibrated to appetite, with automatic routing by appetite and exposure. Commercial levers such as commission, discount bands, and loadings stay bounded by the insurer's own policy, so nothing is priced outside the manual. The insurer remains accountable for the price and the risk.
How long until dynamic pricing goes into production?
Setup runs 3 to 12 months as a fixed-scope implementation, with KPIs agreed before start, after which the platform moves into continuous operation. The sequence covers line selection, integration by API or portal, calibration to the underwriting manual, back-testing against historical submissions and realized losses, and a controlled go-live in shadow or assisted mode. Straight-through authority expands as confidence builds, while human referral stays for out-of-appetite and high-value risks. The actuarial core is never touched.