Case Study

An AI-Native Platform for Insurance Agents and Brokers

OWU24 is Softwise.AI's own AI-native platform for insurance agents and brokers in Poland - consolidating nearly 2,500 General Terms and Conditions (OWU) documents and Product Cards from 40+ insurance companies into a single, continuously updated knowledge base. Built on Retrieval-Augmented Generation (RAG), the platform combines automated document collection, expert-driven categorization, semantic search with citations, and structured product comparison - turning hours of manual document hunting into seconds of natural-language Q&A.

CLIENT

OWU24 (Softwise.AI Product)

YEAR

2025

INDUSTIRES

Insurance

DELIVERABLES

AI-Native SaaS Platform, Automated Document Collection (400+ Sources), RAG-Based Semantic Search, Product Comparison Engine, Expert-Driven Document Categorization

The problem

The Polish insurance market is dense with documentation. Every insurance policy comes with General Terms and Conditions (OWU) and Product Cards - often spanning dozens of pages of fine print - and dozens of insurance companies publish their own versions, refreshed on their own schedules, on their own websites. Insurance agents and brokers must navigate this landscape every day, comparing terms across insurers, keeping up with policy updates, and finding specific clauses on demand to give clients accurate advice.

In practice, doing this manually is brutally slow and risky. The volume is high, the sources are fragmented, the updates are constant, and a single missed clause can mean bad advice - costly for the client and for the agent alike.

Softwise.AI set out to build a product specifically for this market: a single, AI-native knowledge platform that consolidates the entire Polish OWU landscape, keeps itself up to date, and lets an agent get a verified, source-cited answer in seconds.

{Key Challenges}

Why insurance agents lose hours to scattered, ever-changing OWU documents?

The product had to address a tightly interlocked set of structural problems in how insurance documentation works in Poland:

Dispersed sources and large volumes of data
Dozens of insurance companies publish their own OWUs, product sheets, and annexes across separate websites and portals. Manual collection is time-consuming, error-prone, and impossible to keep current at scale - OWU24's knowledge base alone contains nearly 2,500 documents across many insurance lines.

Frequent updates and changes
Insurance policy terms are revised constantly. Without continuous monitoring across many websites and bulletins, agents risk acting on outdated information.

Difficulty finding specific information
Even with every document in hand, finding a single clause - say, whether a home insurance policy covers flooding - is hard. Keyword search demands exact matches, ignores synonyms and context, and forces agents to scan dozens of pages by hand.

Comparing offers across insurers
Clients expect agents to recommend the right product. Comparing two or more policies means analyzing exclusions, coverage limits, territorial scope, and additional clauses scattered across different sections of different documents - tedious to do manually and prone to omissions.

{Process}

How we built OWU24 in 6 stages, from market analysis to live SaaS platform?

Softwise.AI ran the engagement as a productized build, working in close collaboration with an industry expert to make sure every design decision matched how agents actually work.

1. Market and workflow analysis

The team mapped the real daily workflow of Polish insurance agents and brokers - which documents they need, how often they reach for them, what kinds of questions they answer most, and where the comparison decisions are actually made.

2. Expert-driven content categorization

An industry expert defined 13 categories of insurance documents that became the platform's structural backbone, alongside a standardized naming format (Document Type / Insurer / Product Name) - so files are findable both by browsing and by search.

3. Automated collection and monitoring

A dedicated bot was built to scan 400+ sources - insurer websites, industry portals, newsletters - generate change reports, and route new and updated files into the database with full timestamping and archive handling.

4. RAG-based semantic search

Documents were segmented into logical sections (paragraphs), transformed into vector embeddings, and indexed. A language model layer was added on top to generate natural-language answers strictly from retrieved fragments, with citations and links to source documents.

5. Product comparison engine

A dedicated AI tool was built to identify key fragments across multiple products, assign them to relevant comparison categories, and present them in a clear table - each value accompanied by a source quote for full traceability.

6. SaaS delivery and continuous updates

The platform was wrapped in a browser-based SaaS interface for agents and brokers, with daily document updates, in-app weekly change summaries, and monthly email digests keeping the knowledge base evergreen.

Solution

OWU24 is an AI-native SaaS platform delivered via web browser, acting as a central knowledge hub for insurance agents and brokers - combining automated document collection, expert-driven categorization, semantic search, and structured product comparison into a single workflow.

Automatic Document Collection and Updates

A dedicated bot scans more than 400 sources - insurance company websites, industry portals, and newsletters - and generates change reports that drive automatic detection. New and updated files are added daily to a database that now contains nearly 2,000 documents. Each file is timestamped; a Recently Updated section highlights significant changes; users get weekly in-app summaries and monthly email digests; and archived files remain accessible but are excluded from search. Agents stay up to date without manual monitoring.

Slack channel with automated summaries from the web tracker. Each summary groups results into five categories: errors, URLs that could not be processed, modified URLs, URLs that previously errored but are now working, and newly found URLs.

Expert-driven content classification

Every document is classified by insurance type, insurer, and document type, using 13 categories defined by an industry expert. A standardized naming format (Document Type / Insurer / Product Name - for example, Product Card INTER Doctor Real Estate) makes documents easy to locate by name or via a filterable catalog.

Intelligent semantic search with AI (RAG)

The platform's core feature: an AI-powered semantic search engine using Retrieval-Augmented Generation. Documents are segmented, embedded, and indexed. A user asks a question in natural language - "Does insurance X cover flooding?" - and the system retrieves the most semantically similar document fragments, even if phrased differently ("inundation from river water"). Filters (insurance type required, insurer optional) refine the search. A language model then generates a precise, verified answer grounded strictly in the retrieved fragments, complete with citations and links to source documents.

Insurance Product Comparison

OWU24 includes a Product Comparison tool that compares two or more insurance products on the parameters that matter. AI identifies the relevant document fragments, assigns them to comparison categories, and presents a clear table. Each value is accompanied by a quote from the source document - so the comparison is not just fast but fully traceable.

Product comparison tool showing three insurance plans side by side — Allianz (Maximum), Balcia (Maximum) and Euroins (Maximum Plus) — compared across three parameters: rental of a replacement vehicle after an accident, after theft, and after a breakdown.

{Results}

40+ insurers, ~2,500 documents, and one searchable knowledge base for agents

The platform is live and in use by insurance multiagencies; key facts and outcomes include:

40+ insurance companies covered in Poland.
Nearly 2,500 OWU documents and Product Cards consolidated into one knowledge base.
400+ sources scanned continuously for changes - daily updates with no manual monitoring required.
13 expert-defined categories structuring the document estate.
Semantic AI search and product comparison with source citations on every answer.
Daily, weekly, and monthly change tracking so agents always work with the latest data.

{Business impact}

From hours of manual document hunting to seconds of natural-language Q&A

For insurance agents, brokers, and the multiagencies that employ them, OWU24 turns a structural pain point into a competitive edge:

From minutes of search to seconds of answers. Natural-language questions return verified, citation-backed responses - the agent stops scanning PDFs and starts talking to the client.
Always-current knowledge base. Daily monitoring across 400+ sources means agents never advise on outdated terms.
Lower advisory risk. Source-cited answers and traceable comparisons reduce the chance of giving wrong information - protecting both the client and the agent.
Faster client service, more clients served. Time saved per inquiry compounds across the workday, raising service capacity without raising headcount.
A standardised comparison process. Side-by-side product tables with source quotes make recommendation conversations clearer and more defensible.

{Key recommendations}

4 principles for building AI-native vertical SaaS

A few principles emerged from OWU24 that apply to any team building AI products for a regulated, document-heavy industry:

Ground every answer in source documents. In regulated advisory contexts - insurance, legal, financial - citation-backed AI is not a nice-to-have; it is what makes the product safe to use.
Bring in a domain expert for taxonomy, not just for testing. OWU24's 13-category structure and naming convention - defined by an insurance industry expert - is the difference between a searchable platform and a usable one.
Automate the data refresh, or it dies. A document platform whose contents go stale is worse than no platform at all. Continuous monitoring of source sites is part of the product, not infrastructure.
Comparison is the killer feature for advisory tools. In a market with many similar products and complex terms, the ability to put two policies side by side - with quotes - is what changes the buying conversation.

SUMMARY

OWU24 is Softwise.AI's AI-native platform for the Polish insurance market, consolidating nearly 2,500 OWU documents and Product Cards from 40+ insurers into a single, continuously updated, searchable knowledge base. By combining automated document collection, expert-driven categorization, RAG-based semantic search, and structured product comparison - all with source citations - the platform turns the structural pain of insurance documentation into a competitive advantage for the agents and brokers who use it. OWU24 is more than a technological innovation; it is a tangible operational edge in the Polish insurance market.