Elon Musk’s Department of Government Efficiency (DOGE) is reportedly developing a custom AI chatbot for the US government. Named GSAi, this custom generative AI chatbot is designed for the US General Services Administration (GSA). It is part of President Donald Trump’s AI-first agenda to modernise the US government’s operations with the latest technology.
A Wired report cited two sources claiming that the GSAi aims to boost productivity for over 12,000 employees managing office spaces, contracts, and IT services while providing advanced procurement and contract data analysis tools.
The project which has remained largely under the radar, aligns with Trump’s vision to position the US as a global AI leader. Musk’s team has also been accelerating its development, marking a shift from the more cautious AI approach during ex-President Biden’s administration.
What one of the GSAl developers said about the chatbot
In an audio recording (obtained by Wired), Thomas Shedd, a former Tesla employee, now leading Technology Transformation Services (the tech branch of the GSA), mentioned the project in a meeting.
“Another [project] I’m trying to work on is a centralised place for contracts so we can run analysis on them. This is not new at all—this is something that’s been in motion before we started. The thing that’s different is potentially building that whole system in-house and building it very quickly. This goes back to this, ‘How do we understand how the government is spending money?’” Shedd noted.
How this custom chatbot can help DOGE
The GSAi chatbot project by the General Services Administration is expected to help workers by making it easier to write memos quickly. The report cited one of the sources familiar with the project to claim that the agency initially wanted to use existing software like Google Gemini but decided it didn’t meet DOGE’s data requirements.
As per the report, Shedd also stated that deploying “AI coding agents” is one of DOGE’s top priorities. These agents will assist engineers by automatically generating, editing, and answering questions about software code, aiming to improve productivity and minimise errors. The report also cited reviewed documents that reveal that the team explored Cursor, a coding assistant created by Anysphere, which is a startup based in San Francisco.
The report also cited another source familiar with the GSA’s tech purchases who claimed that the agency’s IT team initially approved the use of Cursor but later pulled back for further review. Now, DOGE is reportedly advocating for the adoption of Microsoft’s GitHub Copilot, the most widely recognised coding assistant.
