Generative AI built by KPMG.
KPMG KymChat is a generative AI agent, empowering our people to access the capabilities of ChatGPT in a protected KPMG environment.
Coming together across functions, the KymChat team delivered an AI tool that is unlocking incredible opportunities for our people and our clients.
Based on their experiences with KymChat, the team have a wealth of learning and insight to share.
KymChat - in numbers
* Accurate as of June 2024
Changing the game with Trusted AI: The KPMG KymChat story
From ChatGPT to KymChat
John Munnelly, KPMG Australia’s Chief Digital Officer, recalls returning home from a business trip early in 2023 to a whole new world of AI.
My plane landed and my phone kept buzzing. It was the communications team asking how we should respond to an article about KPMG denying access to ChatGPT. The security team were rightly nervous about the risks and had blocked access for staff. Leveraging our close relationship with Microsoft, we were able to get a secure version of ChatGPT up and running in weeks – we called it KymChat.
John Munnelly
Chief Digital Officer
KPMG Australia
Just as ChatGPT was making headlines, KPMG Australia’s Chief Digital Officer John Munnelly learned Microsoft had acquired 49 percent of OpenAI – ChatGPT’s creator.
With a private connection to ChatGPT, Microsoft offered KPMG Australia early access in a secure environment.
In a matter of weeks, John and the team built a front end for the AI and shared it with KPMG’s senior leadership team. Despite some early known issues with accuracy, it was enough to prove that generative AI was an opportunity worth exploring.
With leadership endorsement, the team built a more robust version of KymChat and deployed the beta less than a month later.
Timeline to launch
November 2022
OpenAI’s ChatGPT is released
January 2023
Microsoft acquires 49 percent of OpenAI
February 2023
KymChat proof of concept is shared with KPMG’s senior leadership team
March 2023
KymChat beta is launched
July 2023
KymChat Personas launched
June 2024
KymChat reaches 10,000 users
Managing risk as KymChat evolves
Collaboration was key to success and involved establishing a multidisciplinary team across technology, risk, legal, security and change management.
During beta, the team ensured all existing data protection measures were applied. John says it was about testing KymChat in a safe environment and putting an explicit ‘beta badge’ across it.
“We tracked every request to KymChat which allowed us to work through any issues. We learned what it was being used for and what gaps we needed to fill.”
Today, the beta sticker is gone and KymChat continues to go from strength to strength. To improve accuracy, the team has created custom libraries of curated datasets, such as the tax precedence database – raising accuracy from 60 percent to 94 percent.
Staff can also rate the accuracy of responses and use citations for verification.
Working smarter with KPMG and generative AI
KymChat is moving the needle on productivity at KPMG, helping staff to:
- Conduct research
- Create first drafts of emails, presentations and documents
- Understand policies.
Jason Fogaty, Partner at KPMG’s Connected Technology Group, says the KymChat strategy is evolving based on learnings captured through each use case and the rapid pace of AI technology development and thinking.
“We have worked out how to step into our service lines to create specialised KymChat products, like our tax offering which helps create the first draft of tax advice for review by our subject matter experts.”
Today, there are several purpose-built versions of KymChat. These are used extensively across KPMG and by our clients, including ASX 100 companies.
Changing the game with Trusted AI
KPMG Trusted AI is our strategic approach to designing, building, deploying and using AI strategies and solutions in a responsible and ethical manner.
Jessica Wyndham, KPMG Australia’s Trusted AI lead, works across the firm to establish governance for the responsible use of AI.
Her starting point is the Trusted AI framework which sets out the principles and pillars that guide KPMG’s responsible development, procurement and use of AI. She says our Trusted AI framework helps the team ask questions like: What are the potential human impacts of what we’re developing?
Every generative AI solution at KPMG needs to be reviewed against the framework as part of the firm’s AI management system. That system includes a robust layer of governance, led by the Trusted AI council.
Nicola Stone, Audit Technology & Innovation Director, explains the importance of our Trusted AI council in delivering responsible AI.
“We have a cross-organisational group of senior stakeholders on the Trusted AI council from across core business units and divisions. We also have something that’s quite unique but vital – an independent external member to provide an objective view.”
AI to empower humans, not replace them
KPMG puts humans at the core of Trusted AI.
John says that AI automates tasks, not jobs, and has created demand for new skills like prompt engineering.
“Every job has several tasks and AI is good at a number of them, but not all. Tasks may be replaced but employment stays the same.”
While generative AI agents like KymChat boost efficiency, human oversight remains critical. Educational initiatives like our ‘24 Hours of AI’ events have helped staff to understand the parameters of responsible AI adoption.
From a culture perspective, KymChat has fuelled a sense of wonder and excitement at KPMG – driving unprecedented innovation and collaboration across the business.
How AI is unlocking opportunities for clients
From tax and compliance to operational decision-making, our AI services and solutions are empowering Australian businesses to solve intractable challenges and unlock new possibilities for growth.
We're helping businesses implement their own version of KymChat through an accelerator solution that combines KymChat use cases, features, lessons learnt and bespoke consulting.
As part of our inaugural KPMG Human Rights Innovation Challenge, we’re also exploring how some of Australia's most impactful not-for-profit organisations can use AI to drive social change.
No matter where you are on the journey, the team can help you unlock opportunities and accelerate value with Trusted AI.
KymChat FAQs
-
How secure is KymChat for my organisation to use? KymChat is designed to leverage the benefits of generative AI while also prioritising security, enabling the use of AI technology without exposing data externally. It is part of KPMG’s Trusted AI framework, which guides the responsible development, procurement, governance and use of AI.
-
What is the first step to get started with KymChat? KPMG can help organisations implement their own version of KymChat through an ‘accelerator’ solution that combines KymChat’s use cases, features and learnings with our consulting services. The process is designed to help organisations gain hands-on experience with generative AI while prioritising data privacy and compliance.
-
Does KymChat have any limitations? Although gen AI can boost efficiency, human oversight will still be critical. KymChat is a powerful tool, and while it can automate tasks, it can’t replace the important human decision-making element.
-
Will generative AI impact the workforce? KymChat is designed to empower people, not replace them. It automates tasks so people can participate in more meaningful work, and creates demand for new skills such as prompt engineering. KymChat can also help foster a culture of innovation and collaboration across an organisation.
-
What’s next for KymChat? KymChat is just the start of KPMG’s gen AI journey. Our KymChat strategy continues to evolve based on the learnings captured through each use case and the rapid pace of AI technology development and thinking.
