Core Concepts
These are the core ideas behind Nairo.
They explain how Nairo should be used, how outputs should be reviewed, and what teams should keep in mind when working with AI in insurance.
Nairo supports expert work
Nairo is designed to support insurance professionals, not replace them.
Users can use Nairo to review documents, ask questions, compare materials, run structured tasks and prepare outputs. Important business decisions should still be reviewed and approved by the right human expert.
This is especially important in underwriting, claims, policy review and audit work, where context, judgment and accountability matter.
Work starts from materials
Nairo works best when the right materials are available in the workspace.
These materials may include:
- Policy wordings
- Underwriting guidelines
- Claims files
- Submission documents
- Internal reference documents
- Spreadsheets
- Uploaded case materials
The quality of the output depends on the quality and relevance of the materials available.
Outputs should be reviewed
Nairo can help users produce summaries, comparisons, answers, checklists, memos and structured reviews.
These outputs should be reviewed before they are used in business workflows.
Users should check:
- Whether the answer is supported by the available materials
- Whether important context is missing
- Whether the output is complete and accurate
- Whether the conclusion fits the organisation’s own rules and judgment
- Whether the output needs correction before it is used
Nairo helps reduce review work. It does not remove the need for review.
Sources matter
Where citations or references are available, users can inspect the documents or passages that informed an answer.
This helps teams understand why an output was produced and makes it easier to challenge, correct or validate the result.
Citations and references do not guarantee that an answer is correct. They help users review the output more effectively.
Materials are workspace-specific
Nairo is designed around customer workspaces.
A workspace contains the materials, users and configuration available to that customer or team.
This means Nairo should be evaluated based on the materials available in the relevant workspace, not as a general AI tool disconnected from the organisation’s own reference materials.
Projects are focused workspaces
Projects are persistent document workspaces.
A Project lets users upload materials and work with the Assistant over those materials in a focused context.
Projects can be used for document sets such as:
- Submission packages
- Policy wordings
- Claims files
- Renewal materials
- Internal review packs
- Audit preparation materials
Submission review is one example of a Project workflow, but Projects are not limited to submissions.
Experts are configured assistants
Experts are AI assistants configured for recurring tasks.
An Expert can include a role, instructions, selected reference materials and review guidance. This helps teams create more consistent patterns for repeated work.
Examples include:
- Underwriting review
- Claims file review
- Policy comparison
- Appetite check
- Audit preparation
Experts support specialist review. They do not replace human approval or internal authority processes.
Actions are repeatable tasks
Actions are structured tasks designed to produce a defined output.
An Action may help produce:
- A summary
- A comparison
- A checklist
- A memo
- A structured review
- A broker communication draft
Actions help teams avoid rewriting the same prompt each time. Outputs should still be reviewed before use.
Decision memory is the direction
Nairo’s longer-term direction is to help insurance teams preserve useful reasoning from their work.
Many teams lose valuable knowledge because reasoning lives in emails, documents, conversations or individual experience. Over time, Nairo aims to help teams capture useful review logic, corrections, decisions and rationale so that knowledge can be reused more systematically.
This is product direction. It does not mean every decision-memory capability is available today. See Decision Memory for more detail.
What these concepts do not mean
These concepts do not mean Nairo makes autonomous decisions.
They do not mean every output is automatically correct.
They do not mean every interaction is audit-grade by default.
They do not replace internal compliance processes, authority limits, contractual terms or professional judgment.
Nairo supports expert review. Governance remains with the organisation.