Putting billions in insurance premium data at their customers’ fingertips

How Jinvoy partnered with Allium Data to solve policy intake, data quality, and customer access.

About Allium Data

Comprehensive insurance premium data, in near real time

Allium Data is an insurtech startup providing comprehensive, near real-time data on in-force premiums, policies, coverages, limits, and deductibles - spanning brokers, carriers, and programs. As a new company with no existing systems, Allium Data partnered with Jinvoy in 2022 to build a customer-facing portal. And as needs grew, that expanded into backend data infrastructure and AI integrations.

~$2B

in insurance premium data

12,000+

entities tracked

1,000s

of hours saved with AI data extraction

The challenge

Three problems, solved one at a time

Allium Data needed more than a single fix - the right solution for each challenge arrived as the need did, not all at once.

Difficult-to-extract source data

Policy data often arrived in hard-to-parse formats, like scanned PDF files, with no consistent structure.

No single point of entry

Policy documents and requests arrived through an email inbox, direct upload, data pipelines, and APIs.

A product customers could actually use

A customer facing portal putting a proprietary database of policy data sourced from 1.4+ million pages of documents - directly at users’ fingertips.

The solution

It started with a portal, then evolved to meet growing needs

Jinvoy built Allium Data’s customer portal first. As new challenges came up over time - parsing scanned documents, handling data coming in from everywhere - the platform grew to meet them.

Capture

Automated intake, built on cloud data pipelines

A set of cloud-based data pipelines bring documents in from wherever they arrive - email inboxes (via an API integration), direct upload, and mailing requests tracked through the mail vendor’s API - each captured automatically, without anyone chasing down a file.

Email inbox
Direct upload
Mail vendor’s API
One intake
pipeline

Public data, loaded on its own pipeline

Employee Benefits 5500 filings and Non-Profit 990 filings are pulled in through a separate ETL process - kept independent from policy document intake, since it’s a different data source on its own schedule.

Operate

Review and publish workflow

An interface where policy documents are uploaded, sent for AI-based data extraction through a GPT-based AI integration, automatically categorized, and reviewed by a person before anything publishes to the client portal - pairing AI speed with human oversight. Policy documents are often lengthy and detail-heavy; automating extraction has saved thousands of hours of processing time.

Upload
AI extraction
Categorization
Human review
Publish

Outbound data on demand

Customers can request data through an API call, or receive it as a flat file on a pre-defined cadence - whichever fits best for how they work.

Visualize

An interactive customer portal

A web app built on a modern SPA framework, giving customers dashboards on premium and entity data, in-file search, and benchmarking against similar insureds - plus request tracking, so they always know when a policy document has been received. The same portal surfaces public data too, including Employee Benefits 5500 and Non-Profit 990 filings, alongside Allium’s own proprietary dataset.

“

Working with Jinvoy felt like having a partner who actually understood what we were trying to build. As our needs changed, the platform grew right along with us - one piece at a time, never starting over. We couldn’t have asked for a better team to build this with.

Michael Rost · Founder, Allium Data

Start the conversation

Have a challenge like this one?

Every project starts with understanding what you’re actually trying to solve.

Contact us
arrow-up icon