Inside KPMG's AI Workforce Strategy: How Agents Are Reshaping Business Operations
Most businesses implementing AI agents struggle with accuracy - but not for the reasons they expect. KPMG's AI Lab Director Lachlan Hardesty reveals how the accounting giant is overcoming these challenges while transforming workforce strategies. Discover their framework for successful AI adoption that's delivering real business impact.
The Unexpected Accuracy Challenge
When businesses implement AI agents, they often assume accuracy issues stem from model limitations. KPMG's experience reveals a more fundamental problem: most organizations have never clearly defined what "accuracy" means for their white-collar workflows.
"Businesses are struggling with accuracy not because of the underlying models," explains Lachlan Hardesty, Director of KPMG's AI Lab. "It's the interpretation of what accuracy means in this world of words, in this world of white-collar work. They've never had to define it in the way you need to define it for an agent."
Key insight: Traditional processes often operate on implicit quality standards that humans intuitively understand. AI agents require explicit definitions of success criteria that many businesses struggle to articulate.
KPMG's Integrated AI Workforce Approach
KPMG's experience implementing AI agents internally has fundamentally changed how they approach digital strategy for clients. Rather than treating AI strategy separately, they now integrate it with workforce planning.
"Traditional digital strategy involves implementing a platform and training the workforce around it," Hardesty notes. "With agents, the AI strategy and digital strategy is now integrated with the workforce strategy. You're actually looking holistically at a workforce and thinking about which components become agents rather than retraining personnel."
This approach represents a significant evolution in how enterprises plan digital transformation. At 4:32 in the video, Hardesty explains how this integrated perspective helps clients make better investment decisions about where to deploy agents versus where to retrain staff.
First Successful Agent Implementations
KPMG's first successful agent implementation using Relevance AI addressed aged care scheduling - a complex process involving multiple stakeholders and shifting priorities. The solution was developed through a hackathon and rapidly deployed to six pilot clients.
"Because Relevance was very low-code, we were able to deliver that solution really seamlessly," Hardesty recalls. "We could assess commercial viability quickly and understand whether it had relevance to a client."
Implementation insight: The aged care scheduling agent demonstrated how quickly solutions could move from concept to pilot when using low-code agent platforms. This rapid iteration capability became a key factor in KPMG's agent evaluation criteria.
How AI Changed Digital Strategy
KPMG's work with AI agents has revealed fundamental shifts in how enterprises should approach digital transformation. The most significant change? Moving from platform-centric strategies to workforce-centric strategies.
"Digital strategy is now about which components of your workforce become agents," explains Hardesty. "It's not just about implementing software and training people to use it. You're designing processes where agents and humans collaborate in new ways."
This evolution requires businesses to:
- Map processes with agent-human collaboration in mind
- Define clear handoff points between agents and staff
- Establish new metrics for hybrid workforce performance
Implementation Framework
Based on their experience implementing agents across multiple industries, KPMG has developed a framework for successful adoption:
Step 1: Process Articulation
Clearly define why the process exists and what successful outcomes look like. Many businesses skip this foundational work in manual processes but it becomes essential when automating.
Step 2: Accuracy Benchmarking
Establish measurable accuracy standards that reflect real business needs rather than technical capabilities.
Step 3: Workforce Integration
Design how agents will collaborate with human staff, including handoff protocols and exception handling.
In summary: Successful agent implementation requires more upfront process definition than traditional automation. The payoff comes in more reliable operations and easier scaling.
Current Business Challenges
According to KPMG's research, Australian businesses show particularly low trust in AI compared to other markets. Hardesty identifies two key factors driving this distrust:
"First, people feel AI will take their jobs in an uncertain economy. Second, everyone has seen AI hallucinate - but they remember the failures more than the successes."
Businesses also struggle with practical implementation challenges:
- Data sovereignty requirements limiting cloud-based solutions
- Concerns about IP protection when using third-party models
- Difficulty moving from proof-of-concept to production
At 12:45 in the interview, Hardesty discusses how KPMG addresses these concerns through enterprise agreements with AI providers and offline deployment options.
Future Predictions for AI Adoption
Looking ahead, KPMG anticipates two major shifts in how businesses will use AI agents:
Agent-to-Agent B2B Communication
"Business-to-business communications will be agent-to-agent within two years," predicts Hardesty. This includes invoices, contracts, and procurement processes where the efficiency and traceability benefits outweigh human involvement.
Sales Process Transformation
The commoditization of outreach through AI agents will force a reinvention of sales processes. "I'm fascinated to see what the sales process becomes when both buying and selling are increasingly agent-driven," says Hardesty.
Future insight: As agent-to-agent communication becomes standard, human involvement will focus on high-value exceptions and strategic decisions rather than routine transactions.
Watch the Full Interview
For deeper insights into KPMG's AI implementation approach, watch the full 22-minute interview with Lachlan Hardesty. At 7:15, he shares specific examples of how agents are transforming aged care scheduling operations.
Key Takeaways
KPMG's experience implementing AI agents reveals several critical insights for businesses considering automation:
In summary: Successful AI agent implementation requires clearly defining processes and accuracy standards upfront. The biggest payoff comes from integrating agents with workforce strategy rather than treating them as standalone tools. As agent-to-agent communication becomes standard, businesses that master this integration will gain significant competitive advantage.
Frequently Asked Questions
Common questions about AI workforce integration
The main challenge businesses face with AI agents is defining accuracy in white-collar work contexts. Unlike traditional processes where accuracy was assumed, AI agents require explicit definitions of what constitutes accurate performance.
Businesses struggle because they've never had to articulate these standards before implementing automation. This represents a fundamental shift in how organizations think about quality control in knowledge work.
- Requires clear success criteria definition
- Demands measurable performance benchmarks
- Necessitates ongoing accuracy monitoring
KPMG integrates AI strategy with workforce strategy rather than treating them separately. Instead of just retraining personnel to use digital tools, they analyze which workforce components can become agents.
This holistic approach represents an evolution in how digital strategies are developed for the agentic world. It requires rethinking traditional organizational structures and process designs.
- Combines digital and workforce planning
- Focuses on agent-human collaboration
- Creates hybrid performance metrics
KPMG's first successful agent implementation was an aged care scheduling solution developed through Relevance AI. The low-code platform allowed them to rapidly deliver six pilots for homecare clients.
This demonstrated how quickly agent solutions could move from concept to real-world implementation when using the right tools. The success helped establish agents as a viable solution for complex scheduling problems.
- Developed through hackathon
- Deployed to six pilot clients
- Proved rapid iteration possible
Businesses are overcoming accuracy concerns by returning to first principles - clearly defining why processes exist and what successful outcomes look like. This foundational work, often skipped in manual processes, becomes essential when automating with agents.
It requires articulating what 'good' means for each process. Many organizations discover they've never formally defined these standards, even for longstanding manual workflows.
- Document process purposes
- Define success metrics
- Establish accuracy benchmarks
KPMG predicts most business-to-business communications will become agent-to-agent within 2 years. This includes invoices, contracts, and procurement processes.
The efficiency, traceability, and fairness benefits will drive adoption, leaving only high-value exceptions for human involvement. This transformation will fundamentally change how organizations interact commercially.
- Automated contract negotiations
- Streamlined procurement
- Traceable transactions
GrowwStacks helps businesses implement AI agent solutions tailored to their operations. Whether you need workflow automation, AI integration, or full digital transformation, our team can design and deploy solutions that align with your strategic goals.
We combine technical expertise with business process knowledge to create implementations that deliver real operational impact. Our approach focuses on measurable outcomes rather than just technology deployment.
- Custom agent development
- Workforce integration planning
- Ongoing performance optimization
Ready to Transform Your Workforce with AI Agents?
Every day without AI integration puts your business at a competitive disadvantage. GrowwStacks can help you implement KPMG's proven framework to automate processes while maintaining accuracy and control.