AI Assistants as Individual as Your Processes

by | 17.04.2025 | using objectiF RPM

The demands on modern requirements engineering are growing, as are the opportunities to break new ground with the right tools. With the latest releases of objectiF RM and objectiF RPM, you can add an AI assistant to your team that not only verifies information, but also thinks for itself. It helps generate new requirements and create complete test cases. This saves time, eliminates redundancies, and breathes new life into your projects.

What exactly is generative AI? How does it work? How will the configured AI assistants in our tools support your daily work in the future? We answer these questions in our blog post, “Generative AI in Requirements Engineering.”

But that’s not where it ends. Starting with version 9.2, you can configure AI assistants for any artifacts and their relations. This expanded feature surpasses the previous focus on particular requirements engineering artifacts. Use our tools to model your domain-specific concepts and tailor your interaction with AI to your needs.

We will explain what an individual configuration can look like using the example of deriving measures from risks.

The AI Assistant as a Virtual Risk Manager

In this scenario, the AI assistant acts as an experienced risk manager. After identifying risks, you can use the assistant to analyze their impact and generate specific measures based on the analysis. The key feature is that: These measures are formulated not as rough recommendations but as specific, implementable requirements for the system to be developed. The AI assistant is designed to generate clear, consistent, and verifiable measures that meet product requirements.

Screenshot KI_Assistent bearbeiten der Beschreibung

What Are the Benefits?

This use of AI provides tangible support for companies that carry out complex projects.:

    • Reduced effort in risk management.
    • Greater consistency in deriving measures
    • Better protection against potential risks
    • Direct implementation of measures within the development process
    • Integrated traceability of risks and measures

These are real competitive benefits, especially in dynamic project environments.

From Risk Analysis to Requirements

The AI assistant operates according to a fixed principle: each measure is formulated as a clear requirement. Consequently, concrete, actionable requirements are directly recorded for the findings from the risk analysis. This establishes a consistent link between risk management, requirements engineering, and implementation. This provides teams with real added value, enabling them to work efficiently and consider risks.

Screenshot KI-Assistent bearbeiten Hinweise

Process Definition of Risks and Measures

In this scenario, “risk” is defined as a user-defined artifact in objectiF RPM. A risk is a potential event that could negatively impact the introduction, use, or maintenance of a product.

A “measure,” on the other hand, is defined as a subtype of the “requirement” stereotype. A subtype is a further specialization of an existing stereotype. In this case, a “measure” is a specific type of “requirement” that focuses on risk minimization. Measures inherit properties from requirements but can also have specific attributes.

A crucial aspect in this context is traceability between risks and measures. This ensures that each identified risk is assigned the appropriate minimization measures and that the effects of changes can be easily evaluated. Risk management is fully integrated into the objectiF RPM project template.

Best Practices for Effective Prompts in Generative AI

The quality of generative AI’s output depends directly on the quality of its input. More precisely, it depends on the quality of the clues, prompts, and context provided. The clearer the AI understands what is expected of it, the better the results.

The art of formulating targeted input prompts is known as prompt engineering. It is becoming an increasingly important skill, especially in complex, technology-driven projects.

What Makes a Good Prompt?

For generative AI to reach its full potential, prompts should adhere to a few basic principles.

  • Specificity

  • General questions lead to general answers. Formulate precisely what is expected in terms of structure, style, tone, or content.
  • Focus on clarity rather than complexity.

  • Avoid nested sentences, double negatives, and unclear wording. AI cannot read minds.
  • Standardize formatting.

  • Use clear structures, such as lists, bullet points, and paragraphs, to make prompts easy to understand.
  • Set clear boundaries.

  • Provide clear guidelines on length, language level, format, and the use of specific terms. This ensures that the answers are relevant and consistent.
  • Context

  • The better the AI understands the technical and content-related context, the more targeted its responses will be. Use configurable context schemas to clearly define relevant artifacts, roles, and relations.
  • Understanding of roles

  • Assign a role to the AI. For example, say, “You are a risk manager in requirements engineering.” This often significantly improves response behavior.
  • Provide target group perspective.

  • Add notes on tone or perspective. For example, address decision-makers in IT or use factual but inviting language.
  • Provide examples

  • Show what a good response should look like using concrete examples. This helps the AI better understand the structure, tone, and intention of the response.
  • Give positive and negative instructions.

  • Formulate not only what the response should contain, but also what is undesirable. This distinction creates clarity.
  • Explain contexts

  • Describe the technical and semantic relationships between the involved artifacts. This enables the AI to process complex tasks much more efficiently.
  • Step-by-step task

  • Complex instructions are easier to process when broken down into sequential steps.
  • Use metaprompts.

  • Tools such as ChatGPT or Gemini can support you by providing suggestions to improve your prompts.
  • Iteratives fine-tuning

  • A good prompt rarely comes about on the first try. Testing, refining, and testing again is worth the effort.

Well-formulated prompts are the basis for productive, AI-supported collaboration, making your work more efficient. With our tips, you can achieve better results with AI more quickly, including with the AI assistants in objectiF RM and objectiF RPM.

 Structured Prompt for the AI Assistant

For our application example, enter the following prompt under “Instructions”:

Perform the following steps:

    1. The risk: Understand the risk in detail by examining its causes and potential effects.
    2. Determine measures: Create specific measures to effectively reduce or completely prevent the risk.
    3. Formulate requirements: Describe the measures in a way that clearly defines the requirements for the system.
    4. Structure the description.
      1. Short introductory sentence: Use active language (subject, predicate, and object) without subordinate clauses and with a maximum of five sentence elements.
      2. Detailed description: Explain how and why the measure reduces or avoids the risk in several sentences.
    5. Assign a short name to each measure that captures the essence of the introductory sentence.
    6. Check effectiveness: Ensure that the proposed measures sufficiently minimize or avoid the risk. If not, add further appropriate measures.

Screenshot KI generierte Risiko Anweisungen als Prompt

Generate Measures for Risks

Next, concrete measures for the identified risk, “data and information leaks,” will be generated with the help of the AI assistant. The figure below provides a structured representation of the risk, including the risk type, probability of occurrence, and description.
Screenshot Risikobeschreibung
The AI assistant view lists the suggested measures based on the risk analysis. These measures can be adopted directly as requirements and implemented in the development process. Measures that have already been adopted are displayed in the lower section of the view.
Integrating risk analysis and measure generation creates a consistent, AI-supported risk management process within the project.
Screenshot der KI-generierten Maßnahmen zum Risiko

Conclusion

Generative AI increases your teams’ productivity and efficiency. With the upcoming versions of objectiF RM and objectiF RPM, you can equip your processes with AI assistants more specifically.
Due to their adaptability with regard to the data model and relationships between artifacts, objectiF RM and objectiF RPM allow you to create added value with your own AI assistants.
Our application example of deriving measures from risks is just one possibility among many, such as:

    • Test cases for use cases,
    • Requirements from change requests or
    • generating tasks from open issues.

Invest time in precisely formulating prompts and hints because the more specific your inputs are, the more relevant and targeted the AI results will be.
Feel free to try out the new version and provide feedback. What works well? How can we improve the assistants’ functionality? We are committed to continuously developing the AI of our tools and look forward to your suggestions.