Make, Buy or Build Together

The new key question in digitalisation

Many manufacturing companies are currently focusing intensively on AI. Expectations are high: fewer downtimes, better quality forecasts, more stable supply chains, and more efficient engineering processes. At the same time, there is a great deal of uncertainty. Which solutions can simply be purchased? Where is it worth developing in-house? And where is a partner needed?

This is precisely where the classic ‘make or buy?’ question falls short. With standard software, this decision was often straightforward: either you bought an established solution and adapted it, or you developed it yourself. With AI, this logic no longer applies.

AI has a profound impact on data, processes and decision-making. In manufacturing in particular, sensitive information is often at stake: machine running times, quality data, design knowledge, maintenance histories and formulations. Anyone who makes the wrong sourcing decision here risks not only high costs, but also a loss of speed, control and differentiation.

The real question is therefore: “Which AI expertise must remain within the company – and where can external solutions or partners usefully accelerate progress?”

Three approaches to AI utilisation and what they really mean

Make AI: maximum control, but high effort

With in-house development, the infrastructure, data assets, ML functionalities and the actual application are entirely under your own control. This is the crucial difference: those who develop in-house retain sovereignty not only over the data, but also over the application itself.

This can make sense if the AI solution is closely linked to proprietary know-how. Take predictive maintenance, for example: if you have your own machinery, your own sensor data and your own fault patterns, you can achieve a level of precision with self-trained models that no off-the-shelf product can match. However, the effort involved is considerable. It requires data engineers, ML engineers, MLOps expertise and clear governance – skills that are in short supply and need to be built up over the long term.

The ‘Make’ approach is suitable when AI is a genuine differentiating factor, sufficient internal data and expertise are available, and long-term control is a priority.

Buy AI: quick to deploy, but with limited customisation

Off-the-shelf AI products or SaaS solutions are often the quickest route to initial results. This can make a lot of sense for standardised processes: document processing, invoice verification, HR automation or traditional office processes can often be covered effectively by existing solutions.

The limitations become apparent where processes become more specific. A generic AI chatbot does not understand specific manufacturing terminology. A standard forecasting tool is not familiar with the particularities of a supplier network. And if sensitive production data must not be transferred to external systems, a pure ‘Buy’ model quickly becomes problematic.

The ‘Buy’ approach is suitable when the use case is standardised, no critical production data is involved, and speed is more important than differentiation.

Build Together: reaching the right solution faster through collaboration

A technology and consultancy partner contributes AI expertise, platform components and implementation experience. The company contributes domain knowledge, data and an understanding of processes. This combination is particularly crucial in manufacturing: even the best technology is of little use if it does not understand the manufacturing context. At the same time, companies do not need to develop every technical component themselves.

This allows companies to retain control over their data and know-how, whilst accelerating implementation. ‘Build Together’ is suitable when AI is strategically important, but internal resources are insufficient to cover development, operation and scaling on their own.

The crucial question: Does the use case have genuine strategic value?

The key criterion for the sourcing decision is not technology, but strategic value. And in the context of AI, strategic value almost always stems from a company’s own data.

A use case will only set a company apart in the long term if it is based on data that no competitor can replicate. Machine data collected over many years, the company’s own quality histories, and company-specific process logic: these are the true raw materials of differentiation. Without this data advantage, competitors will sooner or later develop the same use case themselves, provided it promises a positive ROI. After all, without IP-sensitive data, there is no real protection against imitation.

This leads to a simple decision-making framework: if the strategic value is high and ownership is possible, the company should consider end-to-end in-house development. If full ownership is not possible but a genuine data advantage exists, ‘Build Together’ is the right approach. If the strategic value is low, an external solution is often sufficient – provided the ROI and performance of the available products are satisfactory.

Three typical manufacturing use cases in the sourcing assessment

Predictive maintenance is a candidate for ‘Make’ or ‘Build Together’. The data is company-specific, the fault patterns are unique, and the potential added value is high. Here, a partner can help to move more quickly from existing machine data to robust models.

Supply-chain optimisation often falls into the hybrid category. Existing tools may suffice for standard planning. However, where bespoke supplier networks or sector-specific bottlenecks come into play, greater customisation is required. ‘Build Together’ often makes more sense in such cases.

Administrative automation, by contrast, is often a clear ‘Buy’ candidate. Invoice processing, contract review and HR processes are structured similarly across many companies. The key is to integrate the right solution seamlessly and to set up governance and data protection correctly.

Conclusion: AI sourcing is a strategic question

‘Make, Buy or Build Together’ is not a technical decision. For manufacturing companies, it determines whether AI merely automates individual processes or becomes the new core of value creation in the long term.

The question is not: ‘Which AI solution should we buy?’ but rather: “How do we structure AI so that it delivers rapid results, protects our expertise and sets us apart in the long term?”

Those who answer this question in a structured manner avoid poor investments and lay the foundations for scalable AI. adesso supports manufacturing companies precisely in this regard – from AI strategy consulting and use-case evaluation through to joint implementation on enterprise AI platforms.

Read more about GenAI and Manufacturing Industry.