DOGE has moved its GSAi custom chatbot from a smaller test into wider use at the General Services Administration, putting the tool in front of 1,500 federal workers. The deployment gives a clearer view of how generative AI is being introduced into day-to-day government work, and why the timing is drawing attention.
A chatbot built for general government tasks
GSAi is described as a proprietary chatbot meant to support general tasks. According to the source article, it works through an interface similar to ChatGPT and is tailored in a way that a GSA worker said makes it safe for government use.
The tool is not limited to one model. The default model is Claude Haiku 3.5, while users can also choose Claude Sonnet 3.5 v2 and Meta LLaMa 3.2 depending on the task.
An internal memo presents the chatbot as a broad workplace assistant. It tells employees that they can use the AI-powered chat to draft emails, create talking points, summarize text, and write code. That places GSAi in the same general category as commercial tools like ChatGPT or Anthropic’s Claude, but inside a government setting.
The current rollout follows a February pilot with 150 users within GSA. The product has been in development for several months, and sources told WIRED that new DOGE-affiliated agency leadership accelerated the deployment timeline. DOGE hopes to eventually make the product available across the entire agency.
What workers are being told not to enter
The rollout also comes with explicit limits on what employees should put into the system. One memo warns workers not to type or paste federal nonpublic information into the chatbot. The examples include work products, emails, photos, videos, audio, and conversations meant to be pre-decisional or internal to GSA.
The same warning also covers personally identifiable information. Another memo instructs people not to enter controlled unclassified information.
Those instructions matter because many useful government work tasks involve internal material. A chatbot can help summarize, draft, or organize, but the memo’s boundaries mean workers must decide what can safely be used as an input before they ask for help.
The source article also describes prompt guidance given to employees. The memo contrasts a vague request with a more detailed version. Under ineffective prompts, one example reads: "show newsletter ideas." The effective version reads: "I’m planning a newsletter about sustainable architecture. Suggest 10 engaging topics related to eco-friendly architecture, renewable energy, and reducing carbon footprint."
That example shows the kind of behavior the tool expects from workers: more context, clearer instructions, and a defined output. It also shows that the success of GSAi may depend not only on the model, but on how well employees understand the rules for using it.
Automation arrives during workforce cuts
The chatbot deployment is happening as DOGE continues what the source article describes as its purge of the federal workforce. That connection is one reason the GSAi rollout is being closely watched.
A prominent AI expert who asked not to be named raised the central concern directly: "What is the larger strategy here? Is it giving everyone AI and then that legitimizes more layoffs?" The expert added: "That wouldn’t surprise me."
The concern is not just that a chatbot exists. It is that a tool meant to automate or speed up work is being introduced while parts of the agency are shrinking. In a Thursday town hall meeting with staff, Thomas Shedd, a former Tesla engineer who now runs the Technology Transformation Services, announced that the GSA’s tech branch would shrink by 50 percent over the next few weeks after firing around 90 technologists last week.
Shedd said the remaining staff would work on more public-facing projects such as Login.gov and Cloud.gov. Those projects provide web infrastructure for other agencies. He also said all other non-statutorily required work would likely be cut.
According to meeting notes viewed by WIRED, Shedd told staff: "We will be a results-oriented and high-performance team." The source article also notes that he told staff in early February that he planned to make AI a core part of the TTS agenda.
Beyond GSA, other agencies are watching
GSAi may not remain an isolated experiment. The Treasury and the Department of Health and Human Services have both recently considered using a GSA chatbot internally and in outward-facing contact centers, according to documents viewed by WIRED. It is not known whether that chatbot would be GSAi.
There was also a February project between GSA and the Department of Education to bring a chatbot product to DOE for support purposes, according to a source familiar with the initiative. The engineering effort was led by DOGE operative Ethan Shaotran.
Internal messages obtained by WIRED described GSA engineers discussing a public endpoint that would allow DOE officials to query an early pre-pilot version of GSAI. One employee called the setup "janky" in a conversation with colleagues. The project was eventually scuttled, according to documents viewed by WIRED.
Elsewhere in government, WIRED previously reported that the United States Army is using a generative AI tool called CamoGPT to identify and remove references to diversity, equity, inclusion, and accessibility from training materials.
The unresolved question for federal AI
The GSAi rollout shows a practical version of the federal AI debate. The tool is already in the hands of 1,500 workers. It can help with routine writing, summarization, and code. It also carries strict input warnings and arrives during a period of significant staffing changes.
One employee who used the product gave a measured assessment: "It’s about as good as an intern," adding that it produced "Generic and guessable answers."
That reaction captures the current state of the rollout. GSAi is not being described as a replacement for every kind of work, but it is being positioned as part of a broader push toward AI and automation. The key issue now is how much government work leaders decide to route through tools like this, and what that means for the people already doing the work.