Canada government agency AI tool tackles ‘bottleneck’ of travel and expense inquiries –

An artificial intelligence (AI)-powered tool developed by a Canadian government agency to tackle a persistent ‘bottleneck’ of repetitive staff travel-booking and expense inquiries has achieved a 90 per cent reduction in ‘routine’ inquiries within weeks of its introduction, the agency has said.

The Public Service Commission of Canada (PSC)’s ‘AI Agent for Travel’ is a digital assistant designed to provide public servants with clear and immediate answers on policies and topics such as the National Joint Council Travel Directive (a framework designed to reimburse employees for reasonable expenses incurred while travelling on government business) and financial delegation (the formal authority given to specific managers to approve spending on behalf of the government, within defined limits and rules).

The tool – which was built by public servants, without need for external consultants nor new platforms – is part of a PSC ‘financial management transformation initiative’, launched in 2023 to modernise financial practices and reduce time spent on administrative processing.

The PSC believes that efficiency and cost savings enabled by the tool, which functions in both Canada’s official languages (English and French), will grow ‘significantly’ as it was only rolled out PSC-wide in November 2025.

The agency is now looking to trigger take-up across the Canadian government, telling Global Government Finance this week that the tool had ‘captured the interest’ of more than 30 federal departments and agencies.

JOIN GLOBAL GOVERNMENT FINANCE ON LINKEDIN

Indirect cost savings

Having already achieved a 90 per cent reduction in inquiries, the PSC believes the AI Agent for Travel is on track to achieve $120,000 (about £64,000/ US$88,000) in annual efficiency savings and significant indirect cost savings.

The tool, in theory, enables earlier travel planning and approvals, which in turn allows employees to book flights and accommodations more quickly. In practice, this could translate into hundreds to thousands of dollars in avoided costs per trip, depending on destination and timing, the PSC says. While admittedly modest in isolation, the agency points out that individual trip savings should compound as usage scales.

The tool was initially piloted within the PSC’s finance operations from July to October 2025. During these months the agency’s finance team (comprising a senior manager, manager, financial analyst and accounting officer) used the AI Agent for Travel internally to support client (individual) inquiries, reviewing and validating the information generated by the tool rather than spending hours manually searching policies.

Even at this early stage, the team achieved notable time savings by reducing the need for manual research, the PSC reports. The tool was launched department-wide in November.

Combined with the solution’s scalability, transferability and integration with Microsoft products such as Word, Excel and SharePoint, the agency hopes that the tool’s long-term value will be significantly greater than its initial departmental deployment at the PSC.

‘Plug-and-play’ model

Looked at from a global perspective, the PSC believes the tool demonstrates to public sector finance leaders worldwide that AI adoption ‘doesn’t require high-risk, large-scale overhauls’ and that, instead, ‘narrow, rules-based use cases – such as travel or procurement – offer a low-risk, high-reward entry point into the future of public finance.’

The initiative has been led by PSC chief financial officer (CFO) Farhat Khan and deputy CFO Emile Wandji.

“Our goal wasn’t just to adopt AI – it was to liberate our finance professionals from the ‘routine’ so they could focus on ‘results’,” Khan, who is also the PSC’s vice-president of corporate affairs, told Global Government Finance. “This initiative proves that when finance leaders take the wheel of digital transformation, we can improve both the employee experience and our bottom line simultaneously.”

“By building this in-house, we’ve demonstrated that the public sector already possesses the talent and the institutional knowledge to innovate,” Wandji said. “We didn’t need a massive external contract – we needed a clear problem to solve and the right guardrails to do it responsibly.”

“We designed this with modularity in mind,” said David Lao, PSC’s financial and procurement operations division director. “While it has started with travel, the architecture is a ‘plug-and-play’ model for any complex government policy. We aren’t just solving a travel problem, we’re providing a replicable framework for the rest of the public service.”

RELATED ARTICLE Tax administration ‘particularly demanding’ environment for AI integration: OECD report – a news story (25 September 2025) on a report titled ‘Governing with Artificial Intelligence: The State of Play and Way Forward in Core Government Functions’

‘AI in finance’ use case

The AI Agent for Travel is the first AI agent developed by the PSC, which exists to ‘promote and safeguard a merit-based, representative and non-partisan public service that serves all Canadians, in collaboration with stakeholders.’

Khan and Wandji are due to present the tool to federal small department and agencies CFOs/deputy CFOs on 16 February through Canada’s Small Departments and Agencies (SDA) Finance and Administration Network (FAN).

The former is also due to showcase the tool when speaking during a event (‘AI in Action: Transforming Comptrollership in the Public Sector’) on 25 February organised by the Financial Management Institute of Canada in capital city Ottawa.

As the agency seeks to showcase the tool globally, the PSC is also planning to submit a proposal to a recent OECD ‘global call’ related to ‘Governing with Artificial Intelligence’.

Launched on 20 January, this is an ‘open invitation to governments worldwide to share real-world use cases, policy initiatives and practical tools related to the trustworthy development and deployment of AI across government’. The deadline for responses is 27 February, with selected examples to be featured in the OECD’s ‘AI in Government global repository’ and future reports.