OCFO Code Assist GPT
Department of Homeland Security · FEMA
Overview
- Development stage
- DeployedSource: c) Deployed – The use case is being actively authorized or utilized to support the functions or mission of an agency.
- High-impact designation
- Not high-impactSource: c) Not high-impact
- Impact justification
- Not reported
- Start date
- 2025-04-11 (day precision)Source: 2025-04-11T00:00:00
- Withheld from public reporting
- Not reported
Mission
- Topic area
- Administrative Functions
- Operational functions
- No function tag matched Derived by keyword rules; see methodology
Problem the AI is intended to solve
Code Assist Generative Pre-trained Transformer (GPT) is an internal facing Generative AI (GenAI) tool to augment the FEMA workforce in generating and troubleshooting existing queries in established query languages (e.g., SQL, Java, COBOL). Users enter the language they are querying, and the Code Assist GPT then provides a proposed query based on the elements provided. If the query is unsuccessful, the tool maintains the session, allowing users to prompt for enhancements until expected results are achieved. At the end of the session, all prompts and queries are removed, and no data is stored outside of the active session. The tool provides improves query generation and rapid iteration, saving users 80-90% of the time compared to custom query development. It also supports various computer languages to assist the data analytics community.
Expected benefits
Saving users 80-90% of the time compared to custom query development, rapid iteration which includes error resolution to query complex data sets and multiple computer languages (COBOL, SQL, Python, etc)
System outputs
Leverages Azure Commercial OpenAI within the FEMA system boundary, currently leveraging ChatGPT4o. Code Assist GPT is an internal facing GenaAI tool to augment the FEMA workforce in generating and troubleshooting existing queries in established query languages. The user simply enters the language they are trying to query (i.e. SQL, Java, COBOL, etc) and the Code Assist GPT then provides a proposed query based on the elements provided by the user. If the user experiences an error or the query is not successful, the Code Assist GPT maintains the session as long as the user is still logged in and hasn't restarted the session, the user can continue to prompt the GPT to provide enhancements to the provided query until results are as expected. At the end of the user session, all prompts/queries are removed and no data is stored outside of the active user session.
Technology
- AI classification
- Generative AISource: Generative AI: AI that generates new or synthetic content (e.g., images, videos, audio, text, code).
- System name(s)
- ReadyAI
- Custom-developed code
- Yes
- Public source code
- Not reported
Sourcing
- How it was built
- Contract and in-houseSource: c) Developed with both contracting and in-house resources
- Vendor (as reported)
- Microsoft (OpenAI part of Azure Commercial Cloud offerings)
- Vendors (standardized)
- MicrosoftOpenAI
A vendor is the supplier named by the agency. It does not identify the underlying model or AI technology.
Data and privacy
- Involves PII
- No
- Privacy Impact Assessment
- Not reported
- Authorization to Operate
- Yes
- Demographic features
- Not reported
- Training and evaluation data
- Programming languages were used to test and fine tune the model, such as JAVA Script, COBOL, SQL, Python. The tool leverages the ChatGPT4o model with the inherent capability.The tool was validated with extensive User Acceptance Testing validating inputs and outputs against similar queries that already were written and performed manually. This was then validated with a pilot group of users with an expansion to other user groups for use.
- Federal Data Catalog
- Not reported
Governance
0 of 8 minimum-practice questions answered. A blank answer means the agency reported nothing; it does not mean the practice is absent.
Potential impacts and how they were identified
Not reported