Civic Innovation

Government Digital Services Teams Modeled on USDS and 18F

Digital service teams delivered fast, but lasting change needs procurement power.

Cover illustration for “Government Digital Services Teams Modeled on USDS and 18F”
Cover illustration for “Government Digital Services Teams Modeled on USDS and 18F”

Modeled on the U.S. Digital Service (USDS) and 18F, government digital units recruited outside technologists to reshape how public services are built and bought. USDS launched in 2014 inside the Executive Office of the President, and 18F launched the same year inside the General Services Administration. Both drew on the earlier UK Government Digital Service (GDS). Allied nations such as Australia and Canada attempted similar adaptations, though outcomes varied widely. In 2025 the American teams were dismantled or repurposed under the Trump administration, and new bodies such as Tech Force and Digital Transformation Canada are now testing the model again. While delivering working software shows what is possible, lasting change requires control over purchasing, retained expertise, and steady political backing.

Why USDS & 18F became the template

Launched on August 11, 2014, the USDS changed who handled the government's technology work, going beyond tweaks to contract language. Seasoned technologists, visual specialists, and product leads signed up for temporary stints in public service. Those who balked at permanent government employment could serve a bounded term, bringing commercial expertise straight into the agency.

The USDS sat in the Office of Management and Budget, within the President’s Executive Office, near presidential priorities and able to carry work across agencies, but it could not compel agencies. Its leverage did not rest on command authority. GSA also housed 18F from that year, and the group moved into Technology Transformation Services after its 2016 launch. Because it billed agencies project by project instead of relying on funds appropriated by Congress, it had to keep proving its worth, though it remained somewhat outside yearly budget politics.

Despite these differences, each group embraced iterative agile methods, studied users, favored open source, pooled platforms, and let data guide choices. They also deliberately adopted less hierarchical structures. Together these elements let them replace conventional outsourcing with internal teams, placing specialized skills directly alongside policymakers.

What the model produced before dismantling

Over its decade of operation, 18F contributed to approximately 450 government technology projects. IRS Direct File, the no-cost filing tool it helped create, became a key case for those opposing the closure. The impact reaches past single offerings. Common infrastructure, buying approaches, and open records gave other jurisdictions reusable resources they could modify without building comparable teams themselves.

Procurement reform mattered because it let the people managing public funds apply technical scrutiny. With Waldo Jaquith’s “scrum team years” framework, a software bid could be turned into a measure of team effort. Budget staff could then ask a simple question: did the scope justify the full amount of work proposed? Staff without engineering backgrounds could still challenge the estimate, and 18F steered agencies away from software deals worth millions or even hundreds of millions that failed to deliver.

Jaquith, Robin Carnahan, and Randy Hart wrote the State Software Budgeting Handbook at 18F, which gained readership across Canada and beyond. Professor Amanda Clarke’s IT procurement reform team in Canada drew on 18F’s online writings and materials as well. USDS built unique expertise through a group known as the "procuremenati" that helped agencies purchase software better, a capability unmatched elsewhere in the federal government. These contributions show why partner nations saw the approach as more than a technical fix.

How the UK's GDS set the precedent

The UK’s Government Digital Service came before USDS in several key ways, which makes a simple narrative of American invention and overseas copying inaccurate. Before co-founding USDS, Jennifer Pahlka had already visited GDS, so influence flowed in more than one direction. Its initial Cabinet Office position gave GDS a remit spanning government and institutional influence that later adopters examined, even as USDS teams created their own funding and staffing models.

Academic work published in ACM pointed to both the achievements of this approach across government and where it fell short. The delivery-first, business-borrowed approach brought quick engineering wins, yet whether it could handle deep organizational change and the overhaul of legacy services stayed an open question. Agile bodies inside government, the researchers argued, eventually have to settle into ordinary service-delivery organizations. Mike Bracken, the founding head, put the practitioner's version on departure: "I hoped the system would have had more capacity for institutional reform".

In January 2025, the Central Digital and Data Office became part of GDS, which later joined the Department for Science, Innovation and Technology. In March 2025 the UK government committed £3.25 billion through a Transformation Fund to support digital delivery and emerging technologies, including AI. That investment came with a new review mechanism. In September 2025, GDS introduced its cross-government framework for measuring digital performance across three pilot departments, using custom dashboards plus semiannual Digital Business Reviews.

How Australia and Canada adapted the model

Australia and Canada took clear inspiration from the American case and the earlier UK approach, yet gave their teams varying levels of institutional authority. Australia's Digital Transformation Agency established standards and supervised investment. Canada started with a more limited embedded operation. These decisions determined the scope of each team's potential impact, regardless of how well they executed their technical tasks.

In 2025, the OECD Digital Government Index placed Australia second overall and first for its “digital by design” approach. The Data and Digital Strategy, the Investment Oversight Framework and the Digital Experience Policy drove that result. The last of these set a trustworthy benchmark for how federal services should be delivered. Without genuine oversight mechanisms or practical support, however, adherence relied entirely on each team’s willingness and skill. The DTA never sustained the unified, whole-of-government drive that enabled GDS to steadily raise benchmarks everywhere.

The Canadian Digital Service embraced the principle of short-term tours of duty, but its mandate was narrower than it hoped for. A longtime member of the team described the letdown that followed: "From the start, we had high hopes to change how the government delivered digital services. It didn't work out that way". Canada’s later restructuring widened that mandate and grew the team.

On September 3, 2026, Prime Minister Mark Carney launched Digital Transformation Canada. Shared Services Canada, CDS, and chosen divisions of the Treasury Board of Canada Secretariat, Public Services and Procurement Canada, and Employment and Social Development Canada were folded into this body. This consolidation placed purchasing and infrastructure under one roof. Patrick Pichette, the former Google CFO, became CEO on September 9, 2026, while a fellowship structure enabled industry experts with niche capabilities to serve brief stints. Budget 2026 allocated $1 billion across five years for the Government Transformation Fund, yet no comprehensive organizational performance framework was published.

GSA also housed 18F from that year, and the group moved into Technology Transformation Services after its 2016 launch.

Delivery capacity vs. Institutional reform

A shared obstacle is the gap between how quickly delivery happens and how slowly institutions change. A development team can deliver operational software sooner than public agencies can transform the contractual and administrative structures supporting the service. This work is fundamentally different in kind. Agile delivery enables teams to try and refine a product, yet it grants no authority over procurement practices, institutional culture or existing service arrangements.

Australia’s standards were credible yet weakly enforced, whereas CDS could deliver but lacked a broad enough institutional mandate. GDS demonstrated that central placement in government did not, by itself, remove the obstacles to reform. In the US, 18F used procurement frameworks to influence the system, equipping non-technical officials to ask different questions. A team can shape institutions by circulating know-how and norms, even without formal power of its own.

Short-term specialists add new expertise, but staff rotation alone cannot develop the collective knowledge and political relationships required for large-scale institutional change. Retaining staff serves a different goal than recruiting them. The model requires both service-delivery expertise and institutions capable of retaining and applying the knowledge those staff provide. Teams also require backing from senior political leaders, since effective technical work alone will not give a team the mandate it needs.

How dismantling the USDS exposed political vulnerability

The dismantling of the USDS put that dependence to a direct test. On January 20, 2025, Trump issued an order turning the USDS into DOGE’s U.S. Service. On February 14, the administration dismissed several dozen employees who had joined before the inauguration. On February 25, the 21 staffers still there resigned as a group, saying they would not help weaken core government systems, expose Americans' data or take apart essential public services.

GSA shut down 18F at the start of March 2025, affecting roughly 90 employees. In a 1am email, Thomas Shedd, who heads TTS, called the organization "non-critical". Its cost-recovery model offered no protection from elimination. Neither a lengthy history of delivering projects nor its particular procurement expertise saved it once its executive backers pulled away.

DOGE differed in its goals and in how it imposed control. USDS and 18F, by contrast, partnered with agencies to improve services while leaving agency staff in charge and avoiding new security risks. DOGE made reduction its priority. The later effort retained the same general template for tech-driven involvement, but abandoned the public-minded obligations that had given that model its purpose.

Losing that workforce made returning to the earlier role harder. The administration launched Tech Force only after it had ousted tens of thousands of federal cyber, data, and technology employees whose expertise any incoming group would require. Bringing in new staff offered no immediate remedy for that lost institutional memory. When the temporary DOGE framework ended July 4, 2026, its Service arm shifted focus to public-facing initiatives resembling those of its predecessor. That pivot shows a lasting need for the work, but not lasting backing for the organization handling it.

Tech Force & Digital Transformation Canada compared

Both DTC and Tech Force aim to move beyond the constraints of the earlier framework. The American experiment emphasizes recruitment, so if Tech Force reaches a thousand people it would dwarf the few hundred USDS employed at its height. Yet sheer numbers cannot guarantee the power to drive change. Its focus on junior technical talent is far narrower than the broader mix of roles found within 18F and USDS.

Tech Force’s appointments run two years, briefer than typical tenures at USDS and 18F, which cuts the time available to build agency connections and procurement expertise. Operating at that scale could also make coordination harder within those time limits. Whether expanded delivery capacity can offset a narrower skill mix and a shorter window to learn the institutions where those skills must operate remains uncertain.

DTC pursues the deepest architectural merger yet, combining delivery capacity with standing procurement and infrastructure roles. By keeping its fellowship approach, it preserves the hiring pipeline while avoiding letting transient staff define the entire organization. Appointing Pichette reflects an ambition to operate across the whole of government. Yet neither elevated rank nor structural unification guarantees successful reform, and the absence of a disclosed metrics system keeps oversight in doubt. The available evidence cannot yet determine which initiative more effectively links hands-on execution to lasting structural change.

What the model requires to work

The comparison separates the conditions for entry, mandate and durability. A central position can connect a team to government agendas and interdepartmental choices, but the USDS case showed that proximity is not enough to force cooperation. Australia shows the same limitation when standards stand alone. When backed by enforcement and practical support, a delivery standard can be adopted across the system, but otherwise it rests on what individual teams can do.

Procurement reform ensures that specialist knowledge shapes choices well beyond any single initiative. The budgeting framework and handbook created that link by letting officials question software spending, and their value did not depend on the authors staying around. For experts cycling through public service to have enduring impact, lasting institutions need to retain the expertise, connections, and power required to apply those lessons. That requirement is separate from simply hiring enough staff to build digital tools.

These arrangements depend on political backing, which can also pull them apart. In the 2025 American experience, the record of useful work did not protect a digital team once executive commitment disappeared. What remains to be tested is whether personnel, buying power, responsibility for results and lasting institutional capacity can be combined into a structure sturdy enough to outlast leadership turnover. Can the model build institutions that last as long as the need they answer?

Elodie Reinholt

More in Civic Innovation

← Front page