Rupert Gaster, Chairman of Procurement Heads, discusses with Will Jenkin his recent work focused on helping organisations build sustainable AI capabilities through a practical roadmap for incremental adoption. Will explores the critical decision of whether to build, buy, or partner—providing senior leaders with a clear framework to align AI investments with long-term agility and competitive advantage.
As organisations move beyond the initial hype of generative AI, the focus for the C-Suite has shifted toward building a sustainable, scalable AI Capability. This is no longer merely a departmental experiment but a strategic imperative that influences everything from operational efficiency to customer experience.
However, the path to integrating artificial intelligence into the fabric of a business is fraught with procurement complexities. For senior leaders, the fundamental challenge lies in determining whether to build, buy, or partner. This decision dictates not only the immediate capital expenditure but also the long-term agility and competitive positioning of the enterprise.
The Three Procurement Models
Establishing an enterprise-wide AI capability requires a nuanced evaluation of internal maturity versus external market readiness. There are three primary strategic paths:
1. Developing In-House Capability (The “Build” Model)
This involves recruiting a dedicated internal team of business analysts, data scientists, machine learning engineers (depending on what capability you want to develop), and AI architects to develop bespoke models and infrastructure.
- Pros: This model offers the highest level of strategic control and proprietary advantage. It allows for the creation of intellectual property (IP) that is uniquely tuned to the organisation’s specific data sets and business nuances.
- Cons: The “skills gap” is the most significant hurdle. Competition for AI talent is fierce, and the overhead of maintaining a permanent, high-cost technical department is substantial.
- Risks: There is a high risk of capability stagnation. If the internal team becomes siloed, the pace of internal innovation may fall behind the rapid advancements occurring in the wider AI ecosystem.
2. Engaging Contractors and Consultancies (The “Partner” Model)
Engaging external expertise for project-based delivery or strategic road-mapping provides a flexible, high-impact entry point into AI adoption.
- Pros: Accelerated delivery and access to specialised “tier-one” talent that would be difficult to hire permanently. Consultancies provide an objective view, helping to navigate internal cultural resistance and technical debt.
- Cons: Project-based engagements can be expensive and may lead to a dependency trap where the organisation lacks the internal knowledge to manage the solution once the consultants depart.
- Risks: Knowledge leakage is a primary concern. Without a robust transition-to-operations (TTO) plan, the organisational learning remains with the external partner rather than being institutionalised.
3. Using Specialist AI Vendors and SaaS (The “Buy” Model)
This path involves procuring “off-the-shelf” AI solutions or platforms that offer embedded intelligence (e.g., AI-enhanced ERP or CRM systems).
- Pros: Rapid time-to-value with lower upfront development costs. Responsibility for model retraining, security updates, and infrastructure maintenance rests with the vendor.
- Cons: Limited customisation – your organisation is essentially using the same “intelligence” as your competitors, making it difficult to achieve a unique market advantage.
- Risks: Vendor lock-in and lack of transparency. Many SaaS AI tools operate as “black boxes,” which can pose challenges for auditability and long-term data strategy.
The Critical Technical Foundation
Regardless of the procurement path, an AI strategy will fail without a resilient technical foundation and a strong business drive. Executives must ensure their technical debt is managed before layering AI on top of legacy systems. The following are non-negotiable:
- Data Governance and Sovereignty: AI requires a “Single Source of Truth.” If data is fragmented across legacy silos or is of poor quality, the AI output will be unreliable. Procurement must prioritise solutions that offer robust data lineage and comply with UK GDPR and emerging AI regulations.
- Scalable Infrastructure and Integration: Whether using on-premise hardware or cloud-native environments, the infrastructure must be capable of handling the heavy compute requirements of AI. Seamless integration via APIs into existing workflows is essential to ensure user adoption.
- Ethical Frameworks and Security: As AI becomes more autonomous, the security perimeter changes. Procurement must vet vendors and internal builds for “adversarial robustness” and ensure that bias-mitigation protocols are built into the development lifecycle.
Conclusion: A Procurement Decision Framework
Developing an AI capability is a journey of incremental maturity. To decide which procurement model aligns with your current state, consider this strategic checklist:
- Strategic Intent: Is this AI capability a “commodity” (efficiency-focused) or a “differentiator” (revenue-focused)? (Commodities should be Bought; Differentiators should be Built or Partnered).
- Internal Readiness: Do you have the data maturity to support an in-house build? (If your data is not centralised, start with a Consultancy to fix the foundation).
- Risk Tolerance: Are you prepared for the long-term maintenance of a bespoke system? (If not, SaaS provides the most predictable cost model).
For most public and private sector organisations, the most effective strategy is a Modular Approach: “Buying” standard AI utilities to gain immediate efficiency, while “Partnering” to “Build” bespoke capabilities in areas that offer true competitive advantage – but throughout, you’ll need to be building your own capability in-house to support whatever it is you’re bringing in from the outside and ensure you can be an intelligent customer.
The Author: Will Jenkin has been leading the delivery of IT and Business transformation for 30 years. While his passion lies in turning strategy into reality, more recently he has been focusing on the development of AI capabilities in small and large organisations. This has been achieved through the creation of a blueprint that enables businesses to innovate carefully and progressively along a roadmap for incremental AI adoption.
