How KPMG's AI Assistant Kai Cut Interview Scheduling by 58% and Handled 23,000 Candidate Queries in Year One
When a firm that bills clients for human expertise decides to automate part of its own talent pipeline, the stakes are personal. KPMG — one of the Big Four professional services firms, with over 275,000 employees across more than 140 countries — deployed an AI recruitment assistant called Kai in partnership with conversational AI platform Paradox. Approximately one year after launch, as first reported in April 2025, the results tell a clear story about where AI adds value in recruiting and where it should not.
The Administrative Drag Problem
Before Kai, KPMG's talent acquisition team spent roughly 60 minutes per candidate on interview scheduling alone. That time went to coordinating calendars across multiple interviewers, answering repetitive logistics questions, confirming time zones, and managing back-and-forth communications that rarely required professional judgment. For a firm that hires thousands of professionals annually across audit, tax, and advisory practices, that administrative drag consumed time that experienced recruiters could have spent on candidate assessment, relationship building, and strategic workforce planning.
Sandy Torchia, KPMG's Vice Chair of Talent and Culture, described the motivation plainly: "We weren't trying to build the next robot recruiter; we just wanted to stop wasting human time on things that don't need a human."
The goal was never to replace recruiters. It was to remove the logistics layer sitting between them and the work that actually requires people.
Year-One Results
Within its first 12 months of deployment across KPMG's US recruiting operation, Kai delivered measurable outcomes across four key areas:
- 58% reduction in interview scheduling time, bringing the average from approximately 60 minutes to 25 minutes per candidate. Recruiters retained ownership of candidate selection and assessment — Kai handled only the calendar coordination and confirmation workflow.
- 1,000+ recruiter hours saved across the talent acquisition team. That time was redirected to higher-value activities: sourcing, candidate evaluation, and strategic hiring planning.
- 23,000+ candidate queries handled automatically, covering topics ranging from interview logistics and application status to office directions and benefits questions.
- 21.5% candidate uptake on AI-suggested role recommendations. When Kai identified open positions that matched a candidate's profile, more than one in five candidates actively engaged with those suggestions — indicating that the conversational recommendations were relevant, not intrusive.
These figures reflect year-one performance as reported by Fortune (April 8, 2025) and The Finance Story (April 10, 2025), both citing approximately one year of deployment at the time of publication. KPMG has described the Kai programme as ongoing, though no updated metrics beyond year one have been published as of June 2026.
The After-Hours Advantage
One outcome KPMG did not fully anticipate: 33% of candidate queries to Kai arrived after standard business hours. A third of all candidate interactions would have either gone unanswered until the next business day or required dedicated after-hours staffing under the previous model.
Kai handled these conversations instantly at no incremental cost. For professional services firms with global candidate pools spanning multiple time zones, this kind of always-on responsiveness represents a structural advantage that is difficult to replicate with human-only teams. Candidates in different regions received the same speed of response at midnight as they would at midday — without KPMG adding headcount to cover the gap.
What Kai Does Not Do
KPMG was deliberate about drawing hard boundaries around Kai's role. According to Fortune and The Finance Story, the AI assistant is explicitly excluded from:
- Screening or evaluating candidates — no resume scoring, no ranking, no shortlisting
- Making or influencing final hiring decisions — offers, rejections, and advancement decisions are entirely human-driven
Every consequential decision in KPMG's hiring process remains with a human recruiter or hiring manager. Kai's scope is strictly administrative: scheduling interviews, answering logistical questions, and surfacing relevant role recommendations based on candidate profiles.
This human-in-the-loop design was not an afterthought — it was the founding principle of the entire deployment.
Adam Godson, CEO of Paradox, framed the philosophy: "It's not about automating as much of the process as you can. It's about picking the spots where automation makes sense."
Scaling Beyond the US: KPMG Singapore
Following the successful US rollout, KPMG expanded the Kai deployment to its Singapore operations. KPMG Singapore adopted the same human-plus-AI model, applying conversational AI to administrative tasks while preserving human judgment for all hiring decisions. The Singapore deployment demonstrated that the model transfers across geographies, regulatory environments, and cultural contexts — a key consideration for any multinational firm evaluating AI recruitment tools at scale.
The consistency of results across two very different hiring markets — the US and Singapore — suggests that the administrative bottlenecks Kai targets are not region-specific. Interview scheduling friction and after-hours candidate queries exist wherever large organisations recruit at volume.
A Template for Knowledge-Economy Firms
KPMG's approach offers a practical blueprint that extends well beyond Big Four consulting. The core insight is this: AI recruiting tools deliver the clearest return on investment when they target the administrative substrate of hiring — scheduling, query handling, logistics coordination — rather than the judgment-intensive steps that define candidate selection.
For any knowledge-economy firm where employee quality is the product, the KPMG model demonstrates that AI does not need to touch hiring decisions to transform the recruiting function. Freeing experienced talent professionals from administrative work is itself a strategic gain.
The 1,000-plus hours Kai returned to KPMG's recruiters in year one were not just time saved. They were hours reinvested in the kind of human work that actually requires a human — evaluating potential, building relationships, and making the hiring decisions that shape the firm's future.
What exactly does KPMG Kai AI assistant do?
Kai handles the administrative layer of KPMG recruitment: scheduling interviews, answering candidate questions about logistics and applications, and surfacing role recommendations. It does not screen candidates, rank applicants, or influence hiring decisions.
How much did Kai improve scheduling efficiency at KPMG?
Kai reduced interview scheduling time by 58% — from approximately 60 minutes to 25 minutes per candidate — and saved KPMG talent acquisition team over 1,000 recruiter hours in its first year.
Does Kai make any hiring decisions?
No. KPMG deliberately excluded Kai from candidate screening or evaluation. All decisions about who advances, who receives an offer, and who is rejected remain entirely with human recruiters and hiring managers.
How did Kai handle after-hours candidate queries?
33% of all candidate queries to Kai arrived outside standard business hours. Kai responded instantly at no incremental cost, providing always-on candidate support without requiring additional staffing.
Can other companies replicate the KPMG Kai model?
Yes. The model — deploying AI for administrative tasks while reserving human judgment for consequential hiring decisions — is transferable across industries. KPMG demonstrated this by successfully extending Kai to its Singapore operations after the US rollout.