AI Snippet 17: Requirements Negotiation and Prioritization using LLMs

LLMs can be used for requirements negotiation and prioritization, an innovative approach to improving decision-making in the early stages of development. The example below outlines a methodology for using LLMs as agents representing different stakeholders in a requirements negotiation scenario, ensuring a balanced and well-rounded outcome that aligns with both user needs and technical constraints.??

Example Prompt:

1. Create two agents:

• Agent1 (A1): Represents the primary user <of your system>.
• Agent2 (A2): Represents the system’s software architect (or any other stakeholder).

2. Task: A1 and A2 will negotiate and discuss <REQ_IDs> to determine a priority list for the system's requirements.

3. Agent Focus:

• A1: Focuses on user experience and <insert specific user needs>.
• A2: Considers technical feasibility, integration with existing systems, and the overall architectural perspective.

4. Discussion Dynamics:

The agents can sometimes have differing opinions, leading to a more nuanced and realistic discussion. No decisions should violate the <constraints and non-functional requirements>.

Expected Output Format:

The final output should present:

• The functional requirement IDs in decreasing order of priority.
• A brief rationale for each requirement’s position, based on the negotiation outcomes between A1 and A2.

Utilizing LLMs for requirements negotiation and prioritization streamlines early-stage decision-making and effectively balances user needs with technical constraints. By adopting this approach, teams can achieve more efficient, objective, and well-rounded outcomes, ultimately leading to smoother development processes.

By Chetan Arora and the IREB AI interest group?