Give A.R.I.A. the program you want to launch
A.R.I.A. is the intelligence and orchestration layer around SelectRate. It starts from the Standard Program your organization already has, works your program requirements against it, and shows what already matches, what needs company configuration, what decisions are yours and what it takes to reach a tested configuration. Your team approves and publishes.
A program is hundreds of decisions that have to work together
Building a new program is expert work, and it stays expert work. What makes it slow is not the judgement. It is holding every decision against every other one, in every state, through every version, and proving afterwards that they still line up.
A.R.I.A. carries the connections between them. It reads the record SelectRate governs, lines your requirements up against it, shows what matches and what is missing, separates what the source settles from what only a person can decide, and keeps the reasoning attached to the result.
Launch a program with A.R.I.A.
Start with the program you want to launch and the Standard Program your organization already has. A.R.I.A. works your requirements against what SelectRate governs, shows what is supported and what is missing, prepares the configuration and reads the test results with you, until the program is ready for your team to approve and publish.
Depending on the line, those requirements can cover states, classes and operations, eligibility, coverages, limits and deductibles, loss cost multipliers, modifiers, carrier rules, underwriting requirements, forms and elections, and the effective date. No line uses all of them, and A.R.I.A. works from the ones that apply.
The program you want to build, in your terms: where, for whom, on what paper, at what price, from when.
What SelectRate already governs for that product and jurisdiction, starting from the Standard Program itself: which states it supports, which are ready to rate, which company values each state still needs, the published base edition, the classifications and territories it carries, the filed ranges a carrier positions inside, the versions in force and the dates they took effect.
What the requirements match, what the filed content supports, which states are ready, which values the carrier still chooses, and which questions the source does not answer. Every finding carries the record behind it.
The configuration becomes a new effective dated version in Program Studio, created from the Standard Program and built from the approved decisions. A.R.I.A. prepares that draft through governed interfaces, so the version arrives ready to review rather than ready to start.
The candidate rates a complete risk without publishing anything and without recording a quote. This is the governed test that already gates release: Request to Publish opens only once a rated test has been calculated against the candidate.
What rated, what referred and why, which worksheet step produced an unexpected number, and which of those are configuration gaps rather than filed behaviour. Gaps come back as decisions, with the evidence for each one.
The review step lists exactly what the new version changes against the published configuration, state by state. Your team reads it, resolves what is theirs to resolve, and submits it.
Publication runs under SelectRate's release governance. Every quote, policy and worksheet records the version it was rated with, so a result stays reproducible after later changes.
A.R.I.A. follows the products and jurisdictions SelectRate governs, which is every ISO and AAIS line on the platform. See the lines SelectRate rates.
Knowledge, Analyst, Activity
Three capabilities on one governed platform. Ask it something, give it something to work out, or move approved work through governed configuration and testing.
A.R.I.A. Knowledge
Ask A.R.I.A.A.R.I.A. Knowledge answers questions about your programs from the governed record itself: products and states, classifications, rules and versions, rating inputs, API fields and program configuration. An answer names what was read and when, and says so plainly when the record does not hold it.
A.R.I.A. Analyst
Give A.R.I.A. something to work out.A.R.I.A. Analyst reads source material and governed configuration, assembles the evidence, separates what the source establishes from what needs a decision, and produces structured findings with a proposed plan. It is the deepest capability in the product, and the reasoning foundation for A.R.I.A.
A.R.I.A. Activity
Turn an approved decision into action.A.R.I.A. Activity moves approved work through governed configuration and testing: it carries an accepted finding out through governed interfaces, inside its permissions and on the record, and where a direct route is not the right one it walks a person through the steps instead. Preparing an approved configuration in Program Studio and running it through Program Test is the case it is built around.
From evidence to action
An answer on its own is not worth much in a regulated file. What matters is being able to show, months later, what was read, what was concluded, who decided and what changed as a result. A.R.I.A. is built so that chain holds together.
What that gives you
- Evidence you can open. A finding shows the passage it relied on, and the passage is checked against the document it came from.
- Disagreement surfaced, not smoothed. Interpretation is run more than once, and a material difference becomes a decision for a person rather than an average.
- Corrections that carry forward. A reviewed decision can become an approved rule, used again only within the scope it was approved for.
- Reproducible later. A completed analysis can be reconstructed from what was stored, so a finding outlives the model that helped produce it.
- One record. Question, source, finding, decision and action stay attached to each other, ready for the file.
Specialised insurance intelligence
A.R.I.A. is built for a narrow job done well: read insurance source material and governed configuration, and produce a defensible finding. That narrowness is the advantage. A bounded task can be measured, and a measured task can be trusted with real work.
That is why A.R.I.A. is not a general assistant pointed at insurance documents. It works on a defined task, against a governed record, and its output is a finding with the evidence attached rather than an opinion.
Knowledge and specialisation are different things
This distinction decides how the product behaves. Facts that change belong in the governed record and are read from it when they are needed: rates, rules, versions, classifications, program configuration, documentation and interface details. A.R.I.A. never carries a copy of them. Specialisation is about the analysis step doing its bounded job well on material it has never seen before, which is what a new filing always is.
Built around SelectRate
A.R.I.A. starts here because SelectRate already provides what an intelligence layer needs and rarely finds: a governed environment with versions, effective dates, tests and an audit trail. Clear ownership is what makes the intelligence useful.
| SelectRate governs | A.R.I.A. brings |
|---|---|
| Rates, loss costs and filed tables | An understanding of what they support and where a program fits inside them |
| Rules, classifications and eligibility | Your requirements matched against what is actually available |
| States, jurisdictions and effective versions | Which version applied, and the reason it did |
| Factors, modifiers and filed ranges | The positions the carrier still chooses, named and ready to decide |
| Rating execution and the worksheet | A result explained from the execution itself |
| Program Studio configuration and versions | The configuration prepared and the changes summarised for review |
| Publication and release governance | The release handed to the people who own it, with the evidence attached |
Rating logic has exactly one home. Keeping it there is what lets A.R.I.A. reason about a program without ever becoming a second opinion on the premium.
The architecture is built around governed interfaces rather than one product, so it can extend to other governed insurance systems in time. SelectRate is where it starts.
Intelligence prepares the work. People control the release.
This is the architecture, not a caveat. The value of preparing work automatically comes from the approval being real, and the approval is real because it is held by roles and credentials rather than by good intentions.
A.R.I.A. is designed to be better at finding the questions than at answering them. A new concept the record has never carried, a value the filing leaves open, a source that contradicts itself: these come back named, with the evidence, rather than quietly resolved.
A.R.I.A. and Standard Programs
A.R.I.A. does not invent a rating program from nothing, and it should not. The starting point already exists: every organization on SelectRate has a Standard Program for each governed product, sitting directly on the published bureau foundation, with its supported states and its outstanding company values already stated.
That changes what an intelligence layer is being asked to do. The question is not "build me a program". It is the far more answerable question of what has to change, between a governed program that already rates and the program you actually want to launch.
So you give A.R.I.A. your program requirements. It reads them against the Standard Program, tells you what already matches, what needs company configuration and exactly which value, which decisions belong to your carrier, what the source does not settle, and where your requirements conflict with each other or with the filing. Then it proposes the configuration and a test plan for it.
Start from the standard. Configure what makes your program different.
Standard Program
A governed program on the published bureau foundation, with its states, its readiness and the company values it still needs.
Your requirements
The program you want to launch, in your terms: where, for whom, on what paper, at what price, from when.
A.R.I.A. analysis
What matches, what needs configuration, what is a company decision, what is missing and what conflicts. With the record behind each finding.
Proposed configuration
A draft version created from the standard, prepared through governed interfaces, with the changes summarised for review.
Governed testing
Program Test rates the candidate without publishing anything, and A.R.I.A. reads the results back as decisions rather than as a log.
Human approval
Review and Request to Publish stay human actions. Publication runs under SelectRate release governance.
Common questions
How is this different from an AI assistant over our data?
A.R.I.A. works from the governed record rather than from general knowledge, and it shows what it relied on. Identifiers, dates, versions and inventories are established by code. Interpretation is bounded, checked against the source and run more than once, so a difference between runs becomes a question rather than an average.
Where does rating logic live?
In SelectRate. Rates, loss costs, rules, classifications, eligibility, jurisdictions, effective versions and rating execution stay in the governed environment. A.R.I.A. reads that environment and reasons about it, which is exactly why its answers can be traced.
Who approves and publishes a program?
Your team. Review and Request to Publish are human actions in SelectRate, and publication runs under its release governance. A.R.I.A. prepares the work and brings the evidence to the review.
What does A.R.I.A. do?
A.R.I.A. Analyst produces evidence based analysis of source material and governed configuration, with findings, a structured proposal and a review report. A.R.I.A. Knowledge answers questions from the governed record. A.R.I.A. Activity moves approved work through governed configuration and testing. All three are part of the product, and all three keep the approval with your team.
Which lines and states does A.R.I.A. cover?
A.R.I.A. follows the products and jurisdictions SelectRate governs, so its coverage grows as the platform's does. The current lines and their status are on the rating platform pages.
Can A.R.I.A. publish a program or underwrite a risk on its own?
No, by design. A.R.I.A. analyses, proposes, and prepares and tests configuration through governed interfaces. Accepting a proposal, choosing the values the filed content leaves to the carrier, setting an effective date and publishing a version are decisions held by your team under SelectRate release governance. Underwriting decisions belong to your underwriters.