Slide 1: AI & ASPICE: innovation driver or compliance risk?
AI & ASPICE: innovation driver or compliance risk? Balancing innovation with traceability, reproducibility, and process integrity BC Process Management December 8, 2025 engineering-solutions.aumovio.com
Slide 2: AUMOVIO Engineering Solutions: We are…
AUMOVIO Engineering Solutions An engineering and technology provider. A one stop shop – from the idea to the product. A gateway to high volume automotive products. WE ARE… More than 1,600 experienced engineers & specialists. Developers of technology for future mobility. © AUMOVIO SE BC Process Manag
Slide 3: Solutions portfolio
AUMOVIO Engineering Solutions Solutions portfolio Consulting & specialist support Integration • Agile development and transformation • System integration • Quality management & process • Virtual integration improvement • Vehicle integration & workshops • Data services • Software integration • Simula
Slide 4: Presenters
Presenters AES business center Process Management Rodrigo Freitas Ashish Pullabhotla Principal consultant Consultant rodrigo.freitas@aumovio.com ashish.pullabhotla@aumovio.com © AUMOVIO SE BC Process Management@AUMOVIO December 8, 2025 4
Slide 5: Would you trust AI in an ASPICE assessment?
Would you trust AI in an ASPICE assessment? Exploring the fine line between automation and accountability “If AI can do requirements, design, coding, and testing - what’s left for engineers?” © AUMOVIO SE BC Process Management@AUMOVIO December 8, 2025 5
Slide 6: Would you trust AI in an ASPICE assessment? Automation and accountability
Would you trust AI in an ASPICE assessment? Exploring the fine line between automation and accountability The line between automation and accountability is blurring… Option 1 Option 2 Create SW test Review SW test Create SW test Review SW test specification specification specification specification
Slide 7: ASPICE: the gatekeeper of trust
ASPICE: the gatekeeper of trust The model behind cap. levels, processes, and evidence Goals − Mitigate project/product risk − Identify process improvements Capability Levels 0 Incomplete 1 Performed 2 Managed 3 Established 4/5 … Traceability Supports consistency, enables impact analysis, provides co
Slide 8: Why AI is entering the ASPICE world
Why AI is entering the ASPICE world AI is becoming a necessity in ASPICE-driven development − Software complexity in modern cars − Vehicles are now software-defined platforms − Explosion of features → more data, more tests, more traceability − Manual processes struggle to keep up with the scale − Ti
Slide 9: AI acting in Management Processes
AI acting in Management Processes Example: MAN.3 project management Potential AI applications Predictive scheduling and effort estimation using historical data Resource allocation optimization across parallel projects Early warning signals for milestone slippage Benefits Risks Improved accuracy in f
Slide 10: AI acting in Support Processes
AI acting in Support Processes Example: SUP.1 quality assurance Potential AI applications Automated compliance checks on documents, code, and models Continuous monitoring of process adherence (e.g., monitoring reaction time and process fulfilment) AI-based anomaly detection in QA metrics (defect rat
Slide 11: AI acting in Development Processes
AI acting in Development Processes Example: requirements & design activities Potential AI Applications Extract, classify, and refine requirements from stakeholder inputs or natural language. Detect ambiguities and inconsistencies early using NLP models. Suggest architecture patterns and component de
Slide 12: When is AI evidence acceptable?
When is AI evidence acceptable? Deterministic vs. stochastic outputs in ASPICE assessments − AI adds an extra wrinkle: You’ll find the foundation in ISO/IEC 33002 − If the AI system is “more” deterministic (e.g., (objectivity, repeatability, reproducibility of configured with fixed prompts, data set
Slide 13: Implications for AI-generated evidence
Implications for AI-generated evidence Practical recommendations Document your AI Ensure version and Stability & generation pipeline traceability reproducibility What tool(s) used, Capture information that the versions/configurations, Store generated outputs in a generation can be repeated or prompt
Slide 14: AI limitations in the V-Model lifecycle
AI limitations in the V-Model lifecycle Why engineer remains critical Black-box Lack of Hallucinations Memory & performance decisions domain experience constraints AI outputs often lack No intuition, no tacit engineering AI can invent requirements, tests, Context limits, execution transparency and e
Slide 15: AI in ASPICE: use-case 1
AI in ASPICE: use-case 1 Process SWE.3 software detailed design and unit construction Process step Role Process step Role Create SW detailed design AI agent Create SW detailed design SW developer Review SW detailed design SW Dev. / SW unit tester Review SW detailed design AI agent / SW unit tester I
Slide 16: AI in ASPICE: Use-case 2
AI in ASPICE: Use-case 2 Process SWE.3 software detailed design and unit construction Process step Role Accountability: SW Developer (monitoring)? Create SW detailed AI agent 1 with company guidelines design Review SW Detailed AI agent 2 with requirement, Design architecture, review criteria SW Deve
Slide 17: AI in ASPICE: Use-case 3 requirements elicitation
AI in ASPICE: Use-case 3 Process SYS.1 requirements elicitation Current Approach Exclude irrelevant documents Translate documents into English Convert files into suitable evaluation format MANY OTHER Perform manual review and evaluation APPLICABLE DOCS Derive requirements acc. to standards Compare d
Slide 18: AI in ASPICE: Use-case 3 current AI approach
AI in ASPICE: Use-case 3 Process SYS.1 requirements elicitation Current AI approach Provide all applicable documents to requirements management AI, incl. first level references Prompt for requirements identification/derivation acc. to standards MANY OTHER (e.g., IREB) in a structured, classified, an
Slide 19: AI in ASPICE: the road ahead
AI in ASPICE: the road ahead From opportunity to accountability Outlook Key Takeaways − Regulatory push → ASPICE PAM, and VDA may publish − AI is touching every ASPICE process - opportunity + compliance guidelines/suggestions risk − Tool qualification pressure → AI tools could require qualification
Slide 20: Closing & Q&A
Closing & Q&A From insight to action I hope you had also recognized the AI inconsistency… © AUMOVIO SE BC Process Management@AUMOVIO December 8, 2025 20
Slide 21: Thank you
Thank you Rodrigo Freitas Ashish Pullabhotla Principal consultant Consultant rodrigo.freitas@aumovio.com ashish.pullabhotla@aumovio.com