Unleashing engineering potential with generative AI

Over recent months, companies have intensified their adoption of Gen AI. This, along with Gen AI’s rapid evolution, has led to new practices and roles for engineers

Although generative AI (Gen AI) initially gained recognition in engineering through applications in software development, its scope has broadened to help tackle today’s major engineering challenges. According to Gartner, Gen AI will require 80% of the engineering workforce to upskill through 2027.   

In today’s market context, Gen AI for engineering enables companies to optimize processes, reducing time-to-market by speeding up the production of engineering deliverables and improving product quality and compliance by automating certain quality control tasks. These contributions are especially critical since products and ecosystems are increasingly complex and regulated, with more stakeholders and highly personalized requirements.

In this blog, we explore effective strategies for integrating Gen AI technologies, offer practical recommendations to maximize their impact, and share key insights on how to unlock the value they bring to engineering. 

How to unlock the potential value of Gen AI in engineering  

Companies are struggling to unlock the potential value of Gen AI in engineering. This is not caused by a lack of use case ideas, but rather the lack of an efficient end-to-end assessment supporting the implementation of suitable use cases into productive systems. In addition, companies face difficulties in upscaling their implemented use cases effectively across the entire engineering department. 

Along the engineering value stream, we have been supporting our clients to integrate Gen AI successfully and, more importantly, to maintain profitability sustainably. In this blog, we share our top three lessons to help you reach your own goals with Gen AI. 

Evaluation process 

Choose the right use cases to be implemented by using measurable assessment gateways 

To get the most value from Gen AI over time, it’s important to choose the right use cases and develop a strategic order of pursuit. Many companies try to connect Gen AI’s impact to their KPIs, but they often find it hard because of the overwhelming variety of application options. To act effectively, companies should use goal-oriented evaluation criteria as gateways within the use case decision process. 

To avoid being overwhelmed by too many options, it’s important to use clear and specific criteria that go beyond the simple effort-benefit considerations of how much effort something takes or what benefit it brings.   

As well as incorporating a specific evaluation criteria, task and process silos must be broken in order to optimize complex and interdependent engineering processes. Therefore, we recommend mapping the potential use case to the value stream throughout the entire V-cycle. This helps you evaluate ideas more clearly and see where different parts of the core engineering processes can support each other and create extra value. 

To compare the company readiness with the individual engineering use cases, we have developed an exhaustive assessment method that considers eight dimensions: strategy, governance and compliance, processes, data, IT infrastructure and security, employees, cost and investment readiness, and ethical and ecological impact.  

However, the minimal criteria to be considered within the Gen AI use cases selection process covers four focal points:

• Functional criteria: Business impact of the specific Gen AI use case in engineering. The use case delivers measurable business impact within engineering workflows. 

• Technical criteria: The necessary data and foundational technical requirements are available to implement the use case. 

• Regulatory criteria: The use case complies with legal and regulatory standards, such as the EU AI Act and internal company policies. 

• Strategic criteria: The use case aligns with and enhances the engineering value stream. 

Additionally, it is highly efficient to set a main KPI to ensure the comparability of use cases across the overall engineering Gen AI portfolio. This is an important part of establishing a strategic fit.

Implementation specifications 

Use the full range of relevant data by shifting the focus from engineering text to engineering data 

In markets, we see successfully implemented Gen AI engineering use cases in two key areas: the beginning and end phases of the V-cycle. Exemplary Gen AI use cases to highlight can be found in requirements engineering and compliance demonstration, which are both still highly document-centric and text-based. The most common applications here are conversational agents based on retrieval-augmented generation (RAG) technology. RAG solutions represent one of the most repeatable and transverse applications across the entire value chain, which is why these applications have been at the core of Gen AI strategies for the last few years. 

Both application areas are ideal for starting your Gen AI implementation journey because the solutions are mature, and the results are significant. Our client engagements suggest that by using Gen AI capabilities on technical documentation (e.g., retrieval and summarization), it is possible to generate high efficiency gains and reduce the time engineers spend accessing knowledge and information in the right technical context by up to 50%.

Even though the use of text-based LLMs works very well, the full potential for Gen AI in engineering has not yet been unleashed. Most of the engineering data is not available as a pure text format. Therefore, achieving a higher level of value generation requires overcoming the limitation of a text-based knowledge base. Within engineering, this means including the vast amount of information formats from various data sources across the product lifecycle (e.g., visualizations, diagrams, sensor data, GPS, or even sounds). For application in engineering, we want to highlight the extension of large language models (LLMs) with large multimodal models (LMMs) capabilities. Especially within complex problem definitions, LMMs show a high potential for significant Gen AI usage improvements and operational efficiency across the product development process. We are rapidly discovering the potential transformation of everyday tasks with generative AI for data engineering.

Applying Gen AI 

How to scale up to unlock the full value potential  

Implementation activities of generative AI for engineering are constantly gaining focus. Today, we see companies collecting a full use case funnel, realized in many small implementations of Gen AI addressing specific engineering tasks, resulting in small and often local value gains, respectively. 

Following the rule of “start small, think big,” we share the belief to first gain conviction of the added value and acceptance by implementing such quick wins. Start with simple and cost-sensitive use cases, such as RAG, and progressively extend to more complex use cases. However, we recommend to always keeping the bigger picture of scaling targets in mind. 

An overall AI strategy helps to guide the starting process, to connect existing Gen AI applications, and identifies synergy potentials from the beginning. 

The topic of scaling becomes crucial when using levers to strengthen and expand value generation. A successful upscaling of Gen AI implementations can be executed vertically by expanding the application area or horizontally by linking different Gen AI use cases. As connecting prior local solutions throughout the development process is highly difficult, we want to share the scalability factors we integrate into Gen AI implementation planning and execution. 

Innovation through Gen AI 

Generative AI in engineering is bringing human skills and intelligent automation together to solve complex challenges and shorten the development cycles drastically. The organizations that want to lead in the field of engineering must act decisively, scaling Gen AI strategically to unlock lasting innovation, resilience, and competitive edge.