The AI Dilemma
Implementing AI into a business is a significant challenge, with key factors to consider:
- Data Quality, Security and Model Choice
- Cost of Deployment
- Talent and Skills Gap
- Speed to market and Differentiation
- Risk management
- Control
Data Quality, Security and Model Choice
At the front of the conversation is the reliability of data and trust for AI adoption. The underlying requirement is to have strong data governance, which includes data privacy, security and compliance with all relevant regulations. Fifty-five percent of companies cite data quality as a major barrier to AI adoption, and organisations spend up to 80% of project time just preparing data.
"Data is often an obstacle to the business adoption of AI... Do they trust what the model or algorithm is going to be telling them?"
– Tom Ramjeet – Senior Manager, Data & Analytics
Legacy systems also pose a major roadblock. Many older technologies are incompatible with modern AI, leading to costly updates and increased technical debt. Even when AI is implemented, the results have shown that 30% of models fail to scale due to poor maintenance and integration, impacting ROI.
The choice of model is important too. To ensure the AI outputs are fair, biases must be controlled from the data level and from the model. It is critical to establish data guidelines and frameworks to remove biases and choose or build models that are built from the correct learning models, trained with the right datasets and apply fairness metrics to minimise the impact. Regularly auditing the model's performance would reduce this impact, however this would require expertise and cost.
Cost of Deployment and Competitive Advantage
Businesses must carefully consider whether to build AI solutions in-house or rely on third-party vendors.
"Buy feels like it's more expensive than build, but that's not always the case if there is not a clear strategy in place"
– Rajinder Rai – Director, Tech & Transformation
Building internally requires a company culture ready for trial and error, as success may not be immediate. It would require an operating model shift – looking into an investment in people, processes, data and in house technology infrastructure.
"If you choose to build, you need to be in the mindset of being prepared to test and learn because you won’t get it right first time."
– Tom Ramjeet – Senior Manager, Data & Analytics
On the other hand, buying is a quick way of building capability within your organisation. However, businesses must consider their AI use case and whether it supports core differentiation. Bharat Bhushan, Partner, Data Architecture and Platforms, KPMG noted in an interview, "If it's something that differentiates you... you probably want to own that IP," to develop unique capabilities instead of relying on generic products. He also points out that "if you are dealing with standard off-the-mill items, you wouldn't need to do a custom build," indicating that off-the-shelf AI products can be more efficient for routine tasks.
"If you get a vendor to build a certain feature, you might have an exclusive period... but after that, everybody in the industry has that feature."
– Bharat Bhushan – Partner, Tech & Data
Finally, organisations must weigh the potential loss of competitive advantage that comes with shared features in externally built solutions.
Talent and Skills Gap
"What do you currently have internally? How do you understand what skills you need to have or invest in for your target state?"
– Rajinder Rai – Director, Tech & Transformation
Developing AI solutions requires expertise in machine learning, data science, and software engineering—skills that are in high demand but short supply, with roles growing 74% annually. This talent crunch makes it difficult for companies to recruit and retain the professionals needed to build robust AI systems. More than half of businesses in the UK (51%) acknowledge that they don’t have the right mix of skilled AI talent in-house to bring their strategies to life.
"Organisations that are being seen to be more proactive with adopting AI are offering more exciting career opportunities... they’re going to get the best talent."
– Rajinder Rai – Director, Tech & Transformation
Companies that invest in upskilling existing employees and creating a culture that fosters continuous learning are better positioned to attract and retain top talent. However, to effectively address this skills gap, C-suite executives must conduct a comprehensive assessment of their current talent pool and identify the specific competencies required for their AI initiatives.
Strategic Insights for AI Integration: KPMG AI Strategy
When deciding between building or buying an AI solution, it’s essential to align the decision with the company’s overarching strategic goals and vision. KPMG’s AI Strategy proposition considers a company’s AI journey from a strategic perspective, including value assessment, strategy, and business case development.
Businesses need to apply a decision framework, assessing factors like:
- Core Functionality: What percentage of needs are met by off-the-shelf solutions?
- Configurability: How flexible are the configuration options of potential 'buy' solutions?
- Integration Capabilities: Are essential capabilities accessible via APIs?
- Building Requirements: For bespoke needs, can low-code platforms or APIs be used?
- Team Capabilities: Does the organisation have the right mix of skills to use the available tools effectively?
The Strategy AI proposition guides clients through the essential phases of AI decision-making and business case development.
Step 1: AI Value Assessment
Goal: Identify high-impact AI opportunities
Understanding key GenAI use cases is crucial for how the business can drive growth. This assessment will pinpoint areas within the business capabilities that can yield significant ROI, allowing you to define new strategic objectives that align with growth objectives and set up a path for the future.
Step 2: AI Strategy
Goal: Define your AI vision and objectives
The key aspect here is to clearly articulate an AI vision and ambition through establishing key performance indicators (KPIs). This will lead to outlining future business models from the analysis and prioritisation of AI use cases. Guiding the business toward whether to Buy or Build their AI solution.
Step 3: AI Business Case
Goal: Secure investment and operationalise AI
To gain the necessary approvals and funding, we will provide a detailed cost and timeline estimate for your AI initiatives. This includes a comprehensive implementation plan that addresses dependencies and a strategy for realising the benefits of AI, ensuring you capture maximum value from your investments.
Embracing a More Flexible Future
The build vs. buy decision is no longer a simple choice between two extremes. As technology continues to evolve, particularly in AI, organisations will need to adopt a flexible approach, balancing the benefits of standardised solutions with the need for customisation to meet unique business challenges.
The key for businesses is they must consider the strategic vision and goals to enable their growth through the utilisation of AI.