PLAYBOOK: A roadmap to exponentially scale systems engineering through AI
INTRODUCTION
As systems grow more intricate and evolve at an accelerated pace, organizations face increased challenges managing complexity, ensuring interoperability, and maintaining efficiency across the system life cycle. Traditional engineering approaches struggle to keep up, often resulting in siloed operations, slower development cycles, and increased costs. Model-Based Systems Engineering (MBSE) has emerged as the approach to counter these challenges by shifting from document-based processes to a structured, model-driven methodology. Furthermore, full-scale digital engineering enables organizations to create comprehensive digital representations of systems, allowing for better visualization, simulation and analysis across multiple domains.
While MBSE and digital engineering have advanced system development by integrating cross-discipline models and data, they still rely heavily on human expertise for model creation, interpretation and maintenance. Transitioning to model-centric specifications requires additional training and the quality of implementation can vary based on application, organizational culture and personnel experience. Additionally, maintaining system life cycle data within a company’s ecosystem often demands continuous updates and manual interpretation, limiting scalability and efficiency.
To fully realize the benefits of MBSE and digital engineering at scale, organizations must also leverage advanced technologies like Generative AI (GenAI) to automate critical processes, improve model accuracy and quality, and further optimize engineering workflows. Leveraging GenAI enables human engineers to focus on creativity, experimentation and devising new ways to solve critical problems.
From strategy to deployment, this playbook describes how to make the next technological leap in managing complex systems with AI-enabled systems engineering capabilities. It highlights proven insights to inform and expedite the transition.
WHY GENERATIVE AI FOR MBSE? Examples of benefits and use cases
FORCE MULTIPLIER FOR ENGINEERING
Narrows the gap between the demand for model-based methods and the availability of qualified personnel, saving costs and allowing the engineer to focus on strategic, high-value activities such as innovation and problem-solving
Provides continuous model insights and reasoning with automated reporting of the program’s data and relationships
CONVERT DOCUMENTS TO REUSABLE STRUCTURED MODELS
Aids in knowledge capture and transfer by developing “blueprints” of current and prior system variants that would otherwise not be cost-effective
Accelerates adoption by transforming existing document-based artifacts, such as CONOPS and MIL-STDs, into models
REDUCE RISK IN THE SYSTEMS DEVELOPMENT PROCESS
Supports a more consistent, complete, and correct set of requirements to measure progress
Performs real-time gap analysis to ensure system interoperability
ITERATE FASTER BETWEEN ENGINEERING AND DEVELOPMENT
Generates executable digital environment code or product source code from models quicker and more accurately than ever before
Accelerates systems engineering development time by automatically producing program plans and deliverables
HUMAN CENTRICITY IN GENERATIVE AI
GenAI optimizes human resource expertise by accelerating the analysis of diverse data, enhancing the contributions of subject matter experts (SME) and minimizing time-consuming or repetitive tasks. GenAI handles onerous text processing and tedious tasks (including analyzing disparate information, identifying duplicate information, improving quality, and generating report or text-based artifacts) under the direction of human experts.
AI is not being used to replace human engineers, but rather, to augment their efforts and optimize their expertise; shifting where engineers are focusing their efforts. AI takes advantage of what computers do best—speed, accuracy and natural language processing—while leveraging what human SMEs bring to the table—real-world experience, emotional intelligence and creativity. GenAI can free humans from expending effort on work that doesn’t require their knowledge or expertise, giving engineers the space to innovate and design novel solutions.
AI ADOPTION IN THE SYSTEMS ENGINEERING PROCESS
Augmenting your systems engineering processes with GenAI requires planning, but it doesn’t have to be disruptive. Here’s how to facilitate the transition and set your team up for success.
KEY CONSIDERATIONS
There are three essential steps to implementing AI: identifying where it can help, deciding what processes can and should be AI-supported, and determining which solution to choose, from the many options available. Let’s look at each of these more closely.
IDENTIFY PAIN POINTS — WHERE CAN AI HELP MOST?
Implementing a GenAI-powered engineering solution starts with pinpointing where it can provide the greatest benefits. Consider asking yourself the following questions as a starting point:
What are the repetitive areas in your engineering workflow? Repetitive activities and tasks are time-consuming and pull humans away from higher-level problem-solving. Look to AI to automate these activities. Critical errors are much more likely to occur when humans perform tedious and repetitive tasks, but AI-powered automation can be highly reliable for these processes.
Are there concerns with model quality? Model accuracy and consistency are not inherently achieved through model-based systems engineering and digital engineering. Model maintenance activities are routinely required to ensure good model quality throughout. AI tools can be developed to check the consistency of digital engineering models in real-time and generate dashboard-style data for human engineers to review.
What limits your engineers from achieving their targets and creating more impactful solutions? Humans remain the most important resource. Using AI to relieve humans of the tasks they find boring and mundane improves overall workforce morale and can greatly increase productivity. For example, let AI handle weekly data-based reports so your engineers can innovate and solve problems.
Look to deploy an AI solution to areas in your engineering processes that are mundane, repetitive and time-consuming. Map these pain points to what AI does well, including natural language processing, rapidly processing vast amounts of data, handling multiple processes simultaneously and easily recognizing patterns.
DEFINE THE GENERATIVE AI SOLUTION — WHAT WORK WILL THE AI PERFORM? Based on the above analysis, what processes or tasks can AI support for the humans of the organization? To get the most from an AI-enhanced engineering solution, consider the following:
What data will AI need access to, and what machine steps will it follow to execute the task? By identifying the data inputs and outputs required of your AI solution, you will identify the interfaces and applications. This will help you scope the complexity of the project and narrow down the potential AI solutions.
What are the security concerns? What do you need to protect? AI applications that are not well-secured may inadvertently compromise sensitive information. Determine if your existing access controls and security protocols are sufficient to support the new AI-enhanced workflows.
What data sources do you have available to ground and fine-tune your large language model (LLM)? Making your LLM more accurate and specialized to the task at hand is an important step in enabling an effective and efficient AI solution. The “garbage in, garbage out” adage holds especially true for an AI-powered engineering solution. Understanding the data sources available to ensure your LLM returns more accurate, relevant results will set the trajectory of your AI model’s credibility and coherence.
What are the ethical concerns? Leveraging AI responsibly should always be at the forefront of any AI implementation. Humans need to be aware of biases in the training data as well as biases introduced as work progresses. Even something as innocuous as how prompts are phrased can skew the output. Continuing education and awareness are needed to avoid adding biased information back into the database.
DETERMINE THE RIGHT PLATFORM — HOW DO YOU DEVELOP AND DELIVER THIS CAPABILITY? To support an effective systems engineering workflow, it’s not enough to simply link your engineering tools to a general-use LLM system. While this may decrease deployment time and simplify implementation, it will also dilute the value of the results, because the LLM has been trained on a broad set of general information and may lack technological context. A purpose-built solution will deliver domain-specific responses because it’s been grounded with relevant information.
There is a third hybrid option: leverage a commercial platform with a pre-built domain-specific set of systems engineering capabilities that can be easily integrated with your ecosystem and quickly scaled through the development of new applications to meet emerging needs. Account for success factors — how well will this solution work in your environment, and how technologically relevant are the results.
To achieve a successful implementation, make sure you’ve looked at factors that can influence longterm adaptability and performance, including the following:
Does the platform offer compatibility with other foundation models and LLMs?
Does the platform leverage production-grade Retrieval Augmented Generation (R.A.G.)?
Does the platform offer security and guardrails to protect sensitive data?
Does the platform offer transportability with varied AI platforms and services?
Does the provider offer customizations as a service?
Does the provider offer development tools and/or platforms for customization?
Does it plug and play with open-source AI services?
What visualization or dashboards are available?
Is the platform cost-effective?
What maintenance is required, taking into consideration the security classification of the system?
HIGH-LEVEL ARCHITECTURE
CONSIDER A SKILLED SYSTEMS ENGINEERING TECHNOLOGY PARTNER
A technology partner can often speed up the transition to AI-informed systems engineering, but it’s not always easy to determine who is best suited to provide that support. Ask prospective partners these key questions:
What type of security is built-in (e.g., authentication, role-based access control, etc.), and how will that mesh with our existing security structure?
What out-of-the-box integrations are available (e.g., modeling tool plugins, digital thread automation, web app access)?
Is the solution strictly a pre-trained model integration, or is it purpose-built for the systems?
Does the partner exhibit real-world expertise in MBSE, digital engineering, data, interfaces, and system design?
AI ROLLOUT AND SCALABILITY
To ensure the new system is fully accepted by your organization, it’s essential to consider human factors alongside the technological changes.
Be sure to ask these questions early in the implementation and rollout process:
Is the AI solution modular to enable future growth? Can it adapt to changing business demands and additional users, data and workflows?
How interoperable are the features that you will be leveraging across different AI platforms and repositories (such as Microsoft Azure)?
How will data governance be managed as the use of AI is expanded? How will potential biases be mitigated? Can you ensure transparency about data sources and who has touched the data?
How will the rollout of AI be communicated to the organization, and how will trust be fostered in the solution’s abilities and results?
IMPLEMENTING CHANGE
An incremental rollout allows for time between stages to resolve any issues. This also provides an opportunity to engage team members in understanding the system’s capabilities.
Use pilot programs to implement AI solutions in areas that support highly manual and repetitive parts of the existing engineering workflow
Involve your team in cross-disciplinary learning through collaboration with data scientists and AI specialists
Generate metrics and explainability data to build trust and provide confidence in the solution’s outputs
Ensure training is available from the start with regular updates and reinforcement, especially as new features are introduced
EMPOWER YOUR ENGINEERS WITH ACCURATE, TRANSPARENT, TRUSTWORTHY AI
STC’s AI-informed systems engineering solution was built from the ground up to tackle system designers’ most pressing challenges. Following industry best practices, we create unique graphical user interfaces and workflows to facilitate ethical, explainable and responsible AI.
We can enable AI capabilities across an enterprise or enhance a specific system with solutions that are purpose-built for your environment and your needs. To ensure accuracy and trust, we develop containerized AI applications and solutions (incorporating grounded LLMs, AI agents, retrieval augmented generation, prompt engineering and data cleaning) ready to be deployed in any security domain.
Transparency is also built-in, so you can trace the source of all factors leading to your results as well as who has interacted with the data and system along the way. Our experience and expertise in systems and digital engineering tools, their data structures and application programming interfaces mean we can easily integrate and leverage AI across existing applications, data and workflows.
Our AI infrastructure is secure and scalable to address your evolving challenges. Built with the government in mind, we leverage industry-leading security infrastructure and technology that enable containerization, encryption, regulatory compliance, auditing, identity management and role-based access control. If you have a preferred or proprietary LLM, we can integrate it into a larger capability designed to tackle your most complex projects.
READY TO ACCELERATE AND SCALE YOUR SYSTEMS DEVELOPMENT ACTIVITIES WITH AI?
Learn how STC delivers proven AI-informed systems engineering solutions that save time and deliver better outcomes — whatever the system.