Pennsylvania piloted ChatGPT Enterprise with 175 employees across 14 agencies, who saved an average of 95 minutes daily; the programme scaled to 3,000+ employees across 35 agencies by 2026, shaped by a union-inclusive Labor-Management Collaboration Group.
95 minutes
Average time saved per employee per day
85 %
Participants reporting positive experience
48 %
Participants who had never used ChatGPT before
90 to 60 days
Job description/onboarding cycle time
93 to 34
IT policies consolidated
175 to 3,000+
Employees with generative AI access (2024-2026)
14 to 35
Agencies participating (2024-2026)
6,500
Additional staff enrolled in training
400,000
Documents AI-scanned (DHS)
30 %
Documents flagged for quality review
Details
Maturity
Scaling
Promoter
Commonwealth of Pennsylvania, Office of Administration
Period
January 2024–ongoing (pilot phase concluded May 2025, expanded thereafter)
Keywords
generative AI, digital government, civil service productivity, IT modernization, labor-management collaboration
Context
In January 2024, Pennsylvania's Office of Administration launched a year-long pilot giving 175 employees across 14 state agencies access to ChatGPT Enterprise for tasks such as drafting communications, summarising materials, research and brainstorming. Nearly half of participants (48%) had never used ChatGPT before the programme, and the pilot was co-designed with a Generative AI Labor-Management Collaboration Group including SEIU Local 668 union representatives.
Objectives
To test generative AI for state government productivity while keeping employee and union voices central to adoption decisions.
Activities
Pilot rollout of ChatGPT Enterprise across 14 agencies; official evaluation published March 2025; subsequent expansion of access and training.
Results
Participants self-reported saving an average of 95 minutes per employee per day (roughly eight hours a week), and 85% reported a 'somewhat' or 'very' positive experience. The job-description drafting and onboarding cycle was cut from 90 to 60 days, and 93 IT policies were consolidated down to 34. By mid-2026 the state had expanded generative AI access to more than 3,000 employees across 35 agencies, with 6,500 more enrolled in training; the Department of Human Services used AI-assisted document scanning on roughly 400,000 documents, flagging about 30% for quality review.
Conclusions
Officials, including Governor Shapiro, have stressed that AI functions as a 'job enhancer' requiring human oversight rather than a replacement for staff, and that the tools are not suited to every job type.
Implementation
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
Commonly funded by
National / regional programmes
Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.
Baden-Württemberg built F13, the first AI assistant in productive use across a full German state administration, running on sovereign infrastructure; …
Hong Kong's Digital Policy Office and the HKGAI research centre built HKPilot (document drafting/translation) and HKMeeting (AI minute-taking) for civil …