Qlty uses machine learning including large language models (LLMs) to provide a best-in-class code quality experience. With the power of AI, Qlty is always getting smarter.

This page intends to provide a quick overview of how we leverage these technologies and how we use data for R&D purposes, which varies by service.

Services for individuals and open source projects

For our services for individuals, including our free services for open source projects, we use customer data, including source code, to develop and train models. We have adopted the approach to our personal services based on OpenAI’s own approach to using data from personal ChatGPT accounts for training purposes.

Customer Content from individual services or open source
Is opt-in required?No
Is the underlying data identifiable?Yes
Who will this data (or any output) be shared with?Approved AI subprocessors and other customers
Will this data be used to train Qlty’s models?Yes
Will this data be used to train 3rd party models?No

For more on how we handle data for services for individuals and open source projects, see our Cloud Services Agreement.

Services for businesses

For our services for businesses, we take a restrictive approach. We will not use any elements of your Customer Data that could be intellectual property, personally identifiable information or are otherwise identifiable to a customer (such as source code) without first obtaining opt-in permission.

This will be collected in the form of an in-product opt-in for each separate R&D use case that doesn’t involve Usage Data or Non-Identifying Data. Currently we have two features that use Customer Content, which are AI-generated issue explanations and AI-generated code fixes. These functions must be opt-in enabled for use.

To perform these functions we share the data with approved AI subprocessors, currently OpenAI. Even when these features are enabled for opt-in, neither Qlty nor our 3rd party subprocessors train models with Customer Content.

Customer Content from business services
Is opt-in required?Yes
Is the underlying data identifiable?Yes
Who will this data (or any output) be shared with?Approved AI subprocessors
Will this data be used to train Qlty’s models?No
Will this data be used to train 3rd party models?No

For more on how we handle data for services for businesses, see our Business Terms.

Usage Data and Non-Identifying Data

We use Usage Data and Non-Identifying Data to develop and train AI and ML. Examples of Usage Data would be web analytics and CLI telemetry. Examples of Non-Identifying Data include aggregated data about static analysis issues.

An example of a feature that is powered by Non-Identifying Data is smart prioritization and grouping of static analysis issues. Examples of use of Usage Data are bug fixing, user experience improvements, and A/B testing.

Non-Identifying DataUsage Data
Is opt-in required?NoNo
Is the underlying data identifiable?NoYes
Who will this data (or any output) be shared with?Other customersQlty only
Will this data be used to train Qlty’s models?YesNo
Will this data be used to train 3rd party models?NoNo