How KPMG Is Building AI Agents for Industries with Uniphore
As the World Economic Forum gathers in Davos, KPMG LLP and Uniphore announced a strategic relationship to push AI agents from pilot projects into day-to-day operations. The goal is simple: put governed, high-performing agents inside real workflows to boost accuracy, speed, and outcomes.
The focus starts with procurement and expands across finance, customer experience, workforce optimization, and claims. The through line is enterprise-grade AI that integrates with existing systems instead of sitting off to the side.
What they're building
Uniphore is a Business AI company known for connecting data, fine-tuning models, and deploying agentic AI at scale across large enterprises. In this collaboration, KPMG will build and govern AI agents on Uniphore's Business AI Cloud, paired with small language models (SLMs) adapted to industry context.
The platform integrates with KPMG clients' existing data environments and enterprise systems. That lets teams combine human judgment with AI execution, so agents work inside current processes and controls.
Core capabilities KPMG gains with Uniphore
- Convert regulatory frameworks, institutional knowledge, and process playbooks into industry-specific SLMs for higher accuracy.
- Run governed AI agents across operations like procurement, finance, workforce optimization, claims, and customer experience.
- Support sector-specific use cases in telecommunications, healthcare, financial services, and oil and gas, alongside horizontal solutions.
Under the hood is an "SLM factory" model. Knowledge that lived in documents and people is turned into reusable, governed AI systems that scale.
"We are thrilled to align with Uniphore's vision for AI as a transformative force for business as we focus on helping clients move from AI experimentation to real operational value," says Prasad Jayaraman, Advisory Principal at KPMG.
"Business AI proves its value in production, where enterprise environments are complex, regulated, and deeply interconnected," adds Umesh Sachdev, CEO and Co-Founder of Uniphore. "Our work with KPMG enables a repeatable process for running AI inside real enterprise workflows, so organisations can scale how people and AI work together and drive outcomes."
Early focus: procurement and contracting
KPMG and Uniphore's first wave targets procurement and contracting. Agents classify high-volume contracts, extract obligations, compare terms to approved standards, flag risk, and route exceptions for human approval.
The result: fewer bottlenecks, tighter controls, and less revenue leakage from extended review cycles or missed obligations. Teams get time back to work on negotiations, supplier strategy, and performance.
Built for complex, governed environments
This work is aimed at enterprises with fragmented data, strict governance, and interconnected processes. Agents operate directly within existing workflows and data platforms, preserving current controls and audit trails.
That approach reduces integration friction and makes outcomes measurable. It's a shift away from isolated AI experiments and toward embedded delivery models across global teams.
Why this matters for IT, dev, and product leaders
- Faster path from proof-of-concept to production with governance baked in.
- SLMs tuned to regulations and workflows improve precision and trust.
- Agents are composable and reusable, so wins in one function port to others.
Practical next steps for your team
- Data readiness: map contract repositories, policy libraries, and clause standards. Decide what must stay in-place vs. what can be mirrored.
- Governance: define human-in-the-loop checkpoints, escalation paths, and audit requirements before deployment.
- Model inputs: prioritize the playbooks and regulatory references that SLMs need. Keep source-of-truth ownership clear.
- Pilot scope: pick one high-volume workflow (e.g., NDAs, MSAs, SOWs) with measurable cycle times and risk thresholds.
- Success metrics: track time-to-approve, exception rates, risk findings, and downstream leakage reduction.
- Change management: train reviewers on exception handling and agent feedback loops to improve model accuracy.
What to watch
- Speed of expansion beyond procurement into finance ops and claims processing.
- How industry-specific SLMs reduce false positives/negatives under tight regulation.
- Integration depth with existing CLM, ERP, and data platforms without disrupting controls.
Where to learn more
Leaders from KPMG and Uniphore will discuss their approach at the World Economic Forum Annual Meeting in Davos. For agenda details, see the event hub on the World Economic Forum site.
Build internal capability
If your org is spinning up an AI agent program, upskill your team on practical, production-focused AI. Explore role-based options at Complete AI Training - Courses by Job.
Your membership also unlocks:
