Back office ops · Production

Orizon automates 63% of healthcare code documentation tasks using Databricks GenAI

The problem

Orizon maintained 40,000 medical billing rules coded in legacy languages like C# and C++, adding around 1,500 new rules each month. Each addition required developers to manually document the code and create a flowchart—a several-days-long, error-prone process that bottlenecked business analysts who had to request C++ developers to interpret code.

Workflow diagram · grounded in source
1
New medical rule added
trigger
“Every time a new rule was added, we would have a developer look at the code, document it and create a flowchart”
2
LLMs generate code documentation
ai_action
“their initial use of LLMs for text and graphical documentation of source code saved time and effort, benefiting business users on the product and commercial teams”
3
Business users query rules via Teams
ai_action
“embedding GenAI, along with its natural language processing capabilities, into Microsoft Teams so business users could get answers to questions without relying on developers”
4
Documentation output delivered
output
“the documentation process takes less than five minutes, dramatically accelerating workflows”
Reported outcome

Orizon now processes 63% of tasks automatically, freed up one and a half developers for high-value fraud detection work, cut the documentation process to less than five minutes, and saves approximately $30K per month in better-used resources.

Reported metrics
Tasks processed automatically63%
Monthly cost savings from better-used resourcesapproximately $30K per month
Documentation process timeless than five minutes
Developers freed upone and a half developers
Show all 6 reported metrics
tasks processed automatically63%
monthly cost savings from better-used resourcesapproximately $30K per month
documentation process timeless than five minutes
developers freed upone and a half developers
potential new rules per month (from/to)1,500 to 40,000 per month
potential added productivity1 billion Brazilian reals (BRL)
Reported stack
Databricks Data Intelligence PlatformDelta LakeMLflowDatabricks Model ServingUnity CatalogLlama2-codeDBRXMicrosoft Teams
Source
https://www.databricks.com/customers/orizon
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Orizon now processes 63% of tasks automatically, freed up one and a half developers for high-value fraud detection work, cut the documentation process to less than five minutes, and saves approximately $30K per month…

What tools did this team use?

Databricks Data Intelligence Platform, Delta Lake, MLflow, Databricks Model Serving, Unity Catalog, Llama2-code, DBRX, Microsoft Teams.

What results were reported?

Tasks processed automatically: 63%; Monthly cost savings from better-used resources: approximately $30K per month; Documentation process time: less than five minutes; Developers freed up: one and a half developers (source-reported, not independently verified).

How is this back office ops AI workflow structured?

New medical rule added → LLMs generate code documentation → Business users query rules via Teams → Documentation output delivered.