Organisations Weigh Build, Buy or Partner Options for AI Adoption

United Ceres Workforce Learning has released guidance on how organisations can approach the adoption of artificial intelligence solutions. The framework outlines three pathways: building a solution internally, buying a ready-made product or service, or partnering with an outside provider. It emphasises that the choice should be made for each specific use case rather than applied uniformly across the entire business.

The guidance notes that most organisations use all three approaches for different tasks. It also stresses that regardless of the path chosen, organisations remain accountable for how artificial intelligence is used. The framework references Singapore's Model AI Governance Framework and highlights the importance of vendor and partner checks, information requirements, and managing third-party risks.

How the Three Approaches Work

Building a solution internally means that an organisation develops its own artificial intelligence system from scratch or by customising existing tools. In this scenario, the organisation acts as its own solution provider and retains full control over the design, data, and deployment. This approach can be suitable when the use case is highly specific, when existing products do not meet the requirements, or when the organisation wants to build in-house expertise.

Buying, on the other hand, involves adopting a product or service that is already available on the market. This could be a software platform, a cloud-based service, or a packaged solution that addresses a common business need. Buying can reduce development time and upfront cost, but it also means relying on an external provider for updates, support, and data handling.

Partnering involves working with an outside provider to co-develop or customise a solution. This could take the form of a joint venture, a consulting engagement, or a long-term collaboration. Partnering can combine the organisation's domain knowledge with the provider's technical expertise, but it requires careful management of intellectual property, data sharing, and alignment of objectives.

Why the Decision Should Be Use-Case Specific

The guidance argues that a one-size-fits-all approach to AI adoption is unlikely to succeed. Different use cases within the same organisation may have different requirements for data sensitivity, accuracy, scalability, and integration with existing systems. For example, a predictive maintenance system for factory equipment might benefit from a custom-built solution that integrates with proprietary sensors, while a customer service chatbot might be adequately served by an off-the-shelf product.

By evaluating each use case individually, organisations can match the approach to the specific needs and constraints. This also helps in managing risk: a high-stakes use case, such as a system that makes decisions about loan approvals, may warrant the control that comes with building or partnering, while a low-risk task like internal document search might be fine with a purchased tool.

Accountability and Governance Remain Central

A key point in the guidance is that accountability for AI outcomes cannot be outsourced. Even when an organisation buys or partners for a solution, it remains responsible for how the system behaves, what data it uses, and what decisions it makes. This aligns with the principles of Singapore's Model AI Governance Framework, which sets out expectations for transparency, explainability, and human oversight.

Organisations are advised to conduct thorough checks on potential vendors and partners. These checks should cover the provider's data practices, security measures, compliance with relevant regulations, and track record in similar deployments. The guidance also calls for clear information requirements: organisations should know what data the system will collect, how it will be processed, and what outputs it will produce.

Third-party risks are a particular concern. When a solution depends on external data sources, APIs, or cloud infrastructure, any disruption to those services can affect the organisation's operations. The guidance recommends that organisations assess these dependencies and have contingency plans in place.

The Role of Internal Skills

Even when organisations choose to buy or partner, they still need internal capabilities to manage and oversee the AI solution. This is where AI skills development becomes critical. Teams must understand enough about the technology to evaluate proposals, monitor performance, and intervene when necessary. Without those skills, organisations may struggle to hold providers accountable or to adapt the solution as business needs change.

The guidance suggests that AI skills development should be a continuous effort, not a one-time training event. As AI technology evolves, so too must the organisation's understanding of its capabilities and limitations. This includes not only technical staff but also managers, legal teams, and end users who interact with the system.

Practical Considerations for Each Path

For organisations that choose to build, the guidance emphasises the importance of having the right talent in-house. This means investing in data scientists, engineers, and domain experts who can design, test, and maintain the system. It also means ensuring that the organisation has the infrastructure to support development, such as access to computing resources and data storage.

For those that buy, the focus shifts to vendor selection and contract management. Organisations should evaluate whether the product meets their specific requirements, how it handles data privacy, and what level of support is available. They should also consider the total cost of ownership, including licensing fees, integration costs, and ongoing maintenance.

Partnering requires a different set of skills, including the ability to negotiate agreements, manage joint projects, and align incentives. Trust and communication are critical, as is a clear understanding of each party's roles and responsibilities. The guidance notes that partnerships can be particularly useful when the organisation lacks some internal capabilities but wants to retain a degree of control over the solution.

Looking Ahead

As artificial intelligence becomes more embedded in everyday business processes, the decision of how to acquire and manage these solutions will only grow in importance. The guidance from United Ceres Workforce Learning provides a structured way for organisations to think through their options, balancing speed, cost, control, and risk. By treating each use case on its own merits and maintaining strong governance practices, organisations can adopt AI in a way that is both practical and responsible.