A hiring post tied to Anthony Jancso, cofounder of AccelerateX and one of the earliest known recruiters for Elon Musk’s so-called Department of Government Efficiency (DOGE), points to a new effort to bring AI agents into live federal agency work. The project is being described as a way to automate standardized government processes, with a stated ambition that reaches tens of thousands of federal roles.
The source article frames the effort as part of a broader DOGE push to use artificial intelligence across agencies while also seeking large cuts to the federal workforce. It also shows how quickly the idea runs into practical questions: who controls the work, how the systems would be tested, and whether AI agents are dependable enough for government operations.
A hiring pitch aimed at Palantir alumni
Jancso, a former Palantir employee, posted in a Slack group with about 2000 Palantir alumni that he was hiring for a “DOGE orthogonal project to design benchmarks and deploy AI agents across live workflows in federal agencies,” according to an April 21 post reviewed by WIRED. The source describes AI agents as programs that can perform work autonomously.
In the same post, Jancso said, “We’ve identified over 300 roles with almost full-process standardization, freeing up at least 70k FTEs for higher-impact work over the next year.” The claim suggests a large automation target inside the federal government, though the source also notes that it is not clear who the workers hired for the project would work for.
The post said the roles would be based on site in Washington, DC, and would not require a security clearance. Palantir did not respond to requests for comment.
The reaction inside the Slack group was mixed and often negative. The post received clown face reactions, custom emoji reactions, and some heart reactions. One commenter wrote, “DOGE does not seem interested in finding ‘higher impact work’ for federal employees,” and added, “You’re complicit in firing 70k federal employees and replacing them with shitty autocorrect.”
AccelerateX’s path into government tech
AccelerateX was originally called AccelerateSF. VentureBeat reported in 2023 that AccelerateSF had received support from OpenAI and Anthropic. In its earliest version, the group hosted a hackathon for AI developers focused on using AI to address social problems in San Francisco.
A 2023 Mission Local story said Jancso proposed using large language models to help businesses complete permit forms as a way to streamline construction paperwork and possibly help drive down housing prices. OpenAI did not respond to a request for comment. Anthropic spokesperson Danielle Ghiglieri told WIRED that Anthropic “never invested in AccelerateX/SF,” but did sponsor a 2023 AccelerateSF hackathon by providing free access to API usage while the Claude API “was still in beta.”
In 2024, the venture became AccelerateX and shifted its stated focus. In a post on X announcing the change, the company wrote, “Outdated tech is dragging down the US Government. Legacy vendors sell broken systems at increasingly steep prices. This hurts every American citizen.” AccelerateX did not respond to a request for comment.
According to sources with direct knowledge cited in the article, Jancso disclosed that AccelerateX had signed a partnership agreement with Palantir in 2024. A LinkedIn profile for Rachel Yee, described as one of AccelerateX’s cofounders, appeared to show that the company received funding from OpenAI’s Converge 2 Accelerator. Kay Sorin, another AccelerateSF cofounder, later joined OpenAI several months after the hackathon. Sorin and Yee did not respond to requests for comment.
DOGE’s broader AI push
The hiring effort fits into a wider pattern described in the source. Since its creation in the first days of the second Trump administration, DOGE has pushed AI use across agencies while seeking to cut tens of thousands of federal jobs.
The article lists several examples of AI-related activity connected to DOGE:
- At the Department of Veterans Affairs, a DOGE associate suggested using AI to write code for the agency’s website.
- At the General Services Administration, DOGE rolled out the GSAi chatbot.
- DOGE sought to automate the process of firing government employees with a tool called AutoRIF.
- At the Department of Housing and Urban Development, a DOGE operative used AI tools to examine and propose changes to regulations.
Jancso’s cofounder, Jordan Wick, a former Waymo engineer, has also been active in DOGE, appearing at several agencies over the past few months. Those agencies include the Consumer Financial Protection Bureau, National Labor Relations Board, the Department of Labor, and the Department of Education.
The source also notes that Jancso attended a 2023 hackathon hosted by ScaleAI, and that WIRED found another DOGE member, Ethan Shaotran, attended the same hackathon.
Why replacing 70k roles is disputed
The central technical question is whether AI agents can reliably perform government work across agencies. The article makes clear that experts are skeptical of the scale implied by the hiring post.
A federal employee with knowledge of government contracting, speaking anonymously because they were not authorized to speak to the press, told WIRED: “A lot of agencies have procedures that can differ widely based on their own rules and regulations, and so deploying AI agents across agencies at scale would likely be very difficult.”
That concern is straightforward. Even if two government processes look similar from the outside, each agency may follow its own rules, procedures, and review steps. A system that performs adequately in one workflow may not transfer cleanly to another.
Oren Etzioni, cofounder of the AI startup Vercept, said AI agents can be useful for some work, including using an internet browser to conduct research. But he also warned that their outputs can vary widely and be highly unreliable. The source notes that customer service AI agents have invented nonexistent policies while trying to respond to users.
Etzioni’s point is not that AI has no place in government work. It is that replacing jobs one-for-one is a different claim from making selected tasks more efficient. He said AI can help with certain tasks or improve parts of a workflow, but rejected the idea that it could do the jobs of 70,000 employees. “Unless you’re using funny math,” he said, “no way.”
The unresolved issue is accountability
The article leaves several major questions unanswered because the available facts do not answer them. It is unclear who the hires would work for. It is also unclear how the benchmarks mentioned in the Slack post would be designed, how live workflows would be chosen, and what safeguards would be used if an AI agent made an error.
Those details matter because federal agency work is not just a software workflow. It can involve rules, records, eligibility, public services, and decisions that affect people. The source shows a sharp contrast between a fast-moving automation pitch and expert warnings about reliability, variation across agencies, and the limits of current AI agents.
For now, the clearest fact is that DOGE-linked AI automation is moving from broad ambition toward staffing and deployment language. Whether that becomes a practical tool for federal workflows, or an overextended attempt to replace complex public-sector work, remains the question at the center of the debate.