Roadmap

Nr

R&D

Planned

Status

_{Requirements Verification Planner (RVP)
Creation of a digital assistant to automate requirements analysis and verification planning, assigning verification actions to the appropriate project lifecycle stages.}

_{Requirements Quality Analyser (RQA)}

• _{The RQA tool is currently in the proof-of-concept stage.}
• _{The user experience (UX) is not yet optimized for usability.}
• _{Output is not yet fully structured.}

_{Plannend improvements:}

• _{The output will be more structured and user-friendly.}
• _{Users will be able to pre-configure settings to tailor the analysis to their needs, such as:}
  • _{Selecting which INCOSE rules to apply.}
  • _{Deciding whether requirement types should influence rule selection.}
  • _{Adjusting the strictness of the assessment.}
• _{A test report will be generated to summarize results.}

_{Dataset and functionality:}

• _{The tool leverages a vector database containing key SE documents (see TOC above).}
• _{It can be used both as a digital assistant for drafting needs and requirements, and for assessing the quality of individual statements and sets, based on selected rule subsets.}

_{Chat and knowledge chat}

• _{Users should be able to create Word, Excel, and PDF documents directly from chat conversations.}
  

_{Integration with third party sofware like Relatics}

• _{Goal: enable direct data exchange with Relatics without file-based input/output formats, supporting workflows such as requirements analysis, verification plan creation, and verification report import/updates.}
• _{The platform provides seamless integration with Relatics, removing the need for file-based import/export formats for requirements, verification plans, and verification report updates.}

_{RQA Production based}

• _{The platform provides Rule-based Quality Assurance (RQA), allowing users to apply INCOSE standards and custom rules to requirements, risk and hazard analyses, supported by a unified UX for third-party integrations.}
  

Nr

Issue

Updated with

Planned

Status

_{Scalability platform}

• _{Homepage: The application tile layout did not scale correctly and has been fixed.} 

• _{Enhanced scalability of several applications.}  

_{DRE - readability}

• _{DRE: Chat styling has been improved to enhance readability and usability.}

• _{Improved readability of chat text in DRE.}

_{RQA - issues}

• _{Dutch output review comments were previously too extensive and unreliable. This has been corrected.}
• _{Column titles using mixed languages are now correctly aligned with the selected input language.}

• _{Improved RQA output through an upgrade to a newer LLM model.}
• _{Faster requirement analysis within RQA.}
• _{Implemented English column names in RQA and improved overall robustness and stability.} 

⁰¹⁴

^{DRE - Requirement Formulation Issues}

• ^{Extracted requirements may be incorrect or poorly formulated, particularly in Dutch.}
• ^{Examples include circular or meaningless statements, such as:
  "Projectmanagementplan dient te voorzien in een projectmanagementplan."}
• ^{Process-type requirements are often phrased awkwardly or illogically.}

^{Q3 2025}

^Completed

⁰¹³

^{DRE - user overall experiences}

• ^{DRE 3.0 is currently not user-friendly, slow, and does not consistently deliver the desired results.}
• ^{The English version performs noticeably better than the Dutch version.}
• ^{Dutch output for rewritten or newly created requirements is of insufficient quality.}
• ^{The handling of original text retention is inconsistent.}
• ^{General design and usability adjustments are required.}

^{Q3 2025}

 ^Completed

⁰¹²

^{Chat - link handling}

• ^{Generated URLs open in the same window instead of a new tab.}
• ^{Returning to the chat screen causes loss of context or the entire session.}
• ^{Users expect external links to open in a separate tab by default.}

^{Q3 2025}

^Completed

⁰¹¹

^{Chat - Model Output Differences}

• ^{Users report differences between responses in ChatGPT and in the Hegeman AI assistant.}
• ^{ChatGPT usually provides the correct or more complete answer.}
• ^{This creates confusion and reduces trust in the integrated AI version.}

^{Q3 2025}

^Completed

⁰¹⁰

^{Platform - point issues}

• ^{User Management: Fix user creation, role assignment, and admin controls.}
• ^{Reliability: Improve availability, reduce 504 errors, ensure accessibility on corporate networks.}
• ^{LLM Flexibility: Support multiple or local LLMs.}
• ^{Security & Deployment: Add document security and strengthen app deployment workflows.}
• ^{Unified Design: Implement new UI across all apps and platform modules.}
• ^{SE Knowledge Base: Improve large-scale requirement processing and feedback quality.}
  

^{Q3 2025}

^Completed

⁰⁰⁹

^{DRE - Checks, Balances & Workflow Complexity}

• ^{Users find the number of checks and interactions required to run DRE too high.}
• ^{The overall workflow feels heavy, with too many manual confirmations.}

^Q3 2025

^Completed

⁰⁰⁸

^{Chat History & Session Persistence}

• ^{Users cannot access previous chat history, making it easy to lose work.}
• ^{When clicking a generated URL, navigating back causes the user to lose the full conversation.}
• ^{Users regularly request chat history as a key feature.}

^{Q4 2025}

^Completed

⁰⁰⁷

^{DRE - Drag-and-Drop & Upload Reliability}

• ^{Drag-and-drop does not always work on first use; sometimes only upload via file picker works.}
• ^{The system can freeze or get stuck during document upload, requiring browser restarts.}
• ^{Allocation to the “system of interest” is poor when source documents lack clear structure (e.g., standards documents like Sydney Water).}
• ^{Upload failures can result in Error 504.}

^{Q4 2025}

^Completed

⁰⁰⁶

^{REF - Performance & Robustness}

• ^{REF runs occasionally stall or continue far longer than the predicted runtime.}
• ^{No safeguard exists against accidentally starting overlapping or stacked runs.}

^{Q4 2025}

^Completed

⁰⁰⁵

^{REF - User Interface & Workflow}

• ^{Upload order is not intuitive: requirements appear at the top, documents at the bottom.}
• ^{Application settings/options cannot be locked or saved, forcing repeated re-entry by the user.}
• ^{The REF Requirements Template cannot be downloaded correctly—the link opens inside the current window.}

^{Q4 2025}

^Completed

⁰⁰⁴

^{Chat - UI Display & Layout}

• ^{Certain applications do not display correctly within the chat interface.}
• ^{Users need to scroll excessively, or the interface feels too zoomed in.}
• ^{This impacts readability and overall user experience.}

^{Q3 2025}

^Completed

⁰⁰³

^{REF - Evidence Presentation & Data Consistency}

• ^{proof_doc does not always display a single, correct document reference (filename only).}
• ^{The required concatenation of “[proof_location] [proof_detail]” before Relatics import is currently manual.}
• ^{Evidence is sometimes marked as “accepted” even when it should not be.}
• ^{Dutch verifications do not always follow the verification plan (method & criteria).}

^{Q3 2025}

^Completed

_{DRE - Extraction Quality & Completeness}

• _{Users are unsure whether all relevant content is extracted — missed text fragments reduce trust.}
• _{Traceability to source locations is difficult, leading some teams (e.g., Hegeman) to revert to page-by-page extraction.}
• _{Verbatim extraction often includes text that is not a requirement.}
• _{When extraction quality is unreliable, users prefer selecting “all” + “verbatim” to avoid missing content.}
• _{Additional fields, such as document element and text type, do not deliver useful or accurate results.}
• _{These attributes can introduce noise and reduce clarity rather than improving structure.}

_{REF - Reference Processing & Accuracy}

• _{Location references are sometimes duplicated or inconsistent between proof_location and proof_detail.}
• _{Page numbers in proof_location can be incorrect.}
• _{Multiple REF results per requirement cause complications when updating the correct verification in Relatics.}
• _{Differences in REF outputs between runs on the same dataset.}