↑
Case StudyHealthcare
Healthcare Consulting • United States

Your Documents Hold Answers. Databricks Finds Them

A leading healthcare consulting firm needed to analyze thousands of provider contracts. SunnyData built an AI pipeline on Databricks that reads them automatically, turning weeks of manual review into hours.

See client results
8,000+ Contracts processed in hours Analyzed automatically, at scale.
50+ Data points per contract Extracted automatically.
0 Manual document review required The pipeline does the reading.

Client Challenge

Data locked inside documents of every kind

The firm advises healthcare providers across the US on decisions about hiring, retention and pay. One of its clients needed its provider contracts analyzed to inform a strategic decision.

Each contract held 50+ data points, from base salary and bonuses to non-compete clauses. They arrived as PDFs, Word files or scanned images, often with amendments and separate compensation plans. Reviewing thousands by hand was too slow, and different people read the same terms differently.

The answer had to understand context, handle every format and stay accurate for high-stakes pay decisions.

50+ data points
Illustrative

The Solution

From paperwork to structured data

SunnyData built an AI pipeline on Databricks that reads every contract
and turns it into data consultants can use right away.

Read every format

PDFs, Word files and scans are converted into readable text.

Understand the context

AI models know when an amendment overrides the original terms, when a plan covers several providers, and the difference between salary and bonuses.

Process at scale

Databricks runs thousands of contracts in parallel and tracks model performance, so accuracy keeps improving.

Deliver ready-to-use data

Clean, structured data reaches consultants and dashboards, ready for regional and specialty comparisons.

What it runs on
Databricks Runs the whole workflow and the parallel processing.
MLflow Tracks model versions and performance.
Amazon Textract logo
Amazon Textract Reads documents in any format.
Amazon Bedrock logo
Amazon Bedrock AI models that understand contract context.

From Contract to Data

Follow one contract through

Follow one contract through

Four stages, no manual review

Illustrative representation. Not client data.

The four stages: 1, Arrive: a contract in any format. 2, Read: every page, amendments included. 3, Understand: salary, bonuses and clauses found. 4, Answer: structured data, ready to explore.

Key Benefits Achieved

Business results, before and after

MetricBeforeAfterChange
Contract analysis Weeks of manual review Hours, automated Weeks → hours
Consistency Terms read differently by different people The same rules, every time Consistent
Document formats Handled by hand, one by one PDFs, Word files and scans handled together One pipeline
Consultant time Spent on data entry Spent on advisory work More strategic
Client questions Delayed insights Immediate answers Faster decisions

What they can do now

Faster answers, better advice

Answer clients right away

Decisions that needed weeks of analysis can now happen in real time.

Spend time on advice

Consultants focus on strategy instead of data entry.

See trends at a glance

Regional benchmarks and specialty trends are ready to explore, including in Databricks dashboards.

Ready to turn your documents into data?

The same approach works for contracts, claims, invoices and any document your teams still read by hand.

Chris Reid
Chris Reid Account Executive, Healthcare & Life Sciences
Are your teams still reading documents one by one?