Workshops: Agentic Coding for Roles & Industries

Agentic Coding in Automotive Software: ASPICE & ISO 26262 Workshop

Automotive software teams work in processes where every line of code must trace back to a requirement and every tool must earn confidence. Coding agents can still help with tests, documentation, codebase analysis and non-safety components. This workshop shows OEM and supplier teams how to use Claude Code, GitHub Copilot or Cursor within ASPICE and ISO 26262 without putting assessments or the safety case at risk.

Who is this workshop for?

  • Software developers and architects at OEMs and suppliers
  • Functional safety managers and engineers
  • ASPICE process owners, quality managers and internal assessors
  • Teams working with AUTOSAR Classic or Adaptive
  • Test and validation engineers for ECU and vehicle software
  • Engineering service providers in the automotive industry

What you will take away

  • Map agent-assisted work onto the ASPICE software processes from requirements to verification.
  • Maintain bidirectional traceability and consistency when agents generate code, tests or documentation.
  • Assess coding agents as software tools under ISO 26262, including tool impact and error detection.
  • Separate ASIL-relevant components from areas where agents can work more freely.
  • Clarify confidentiality, OEM requirements and cybersecurity before agents touch customer code.

Agenda

  1. Module 1

    Process and safety frame

    • ASPICE: software processes, work products and what assessors look for
    • ISO 26262: ASIL classification and its consequences for development and verification
    • Cybersecurity context: ISO/SAE 21434 and the UN regulations on cybersecurity and software updates
  2. Module 2

    Where agents fit, and where they don't

    • Good candidates: test cases, test harnesses, documentation, tooling scripts, QM-rated components
    • Careful territory: ASIL-rated code, safety mechanisms, timing-critical paths
    • Exercise: sort your backlog by safety relevance and agent suitability
  3. Module 3

    Tool confidence for coding agents

    • Tool impact and error detection: deriving a tool confidence level
    • Controls that reduce tool risk: review, static analysis, coverage, MISRA checks
    • Documenting the tool use case for the safety case
  4. Module 4

    Traceability in practice

    • Linking requirements, architecture, code and tests in ALM and requirements tools
    • Agents draft trace links and review comments, engineers approve
    • Exercise: implement a requirement in an AUTOSAR-style C component with full trace
  5. Module 5

    Supplier reality

    • OEM confidentiality clauses and code that must not leave the company
    • On-premise or EU-hosted model options
    • A rollout plan for one project before scaling

Agents in safety-relevant processes

ISO 26262 treats software tools as potential sources of error and asks how likely it is that a tool error goes undetected. A coding agent is no exception. Code that passes review, static analysis, MISRA checks and requirement-based tests leaves far less room for undetected errors than code that flows straight into a build.

ASPICE adds the process view: assessors look for consistent work products and traceability, regardless of who or what produced them. We show how to keep both intact while agents contribute productively.

Typical rules we work out with teams

These rules are drafts for your process owners; the binding decision stays with your functional safety organisation.

  • Agents only in approved environments, never with OEM data in consumer tools
  • Changes to ASIL-rated code get the same reviews and analyses as hand-written code
  • Every agent contribution is linked to a requirement or change request
  • Tool use cases are documented and re-assessed when tools or models change

Setting and practice environment

The workshop runs remotely or on-site, in German or English, with neooutput AI Code Engineers and AI Solutions Architects. Exercises use a sandbox repository with C/C++ and AUTOSAR-style structures or, if you prefer, approved parts of your own code inside your infrastructure.

Frequently asked questions

Do we have to qualify a coding agent like a compiler?

Not necessarily. Under ISO 26262, the required confidence depends on the tool's impact and on how well errors are detected downstream. With strong verification the required tool confidence level can be low, but your safety organisation must document and accept that rationale.

Can agents write MISRA-compliant code?

With clear instructions and feedback from analysis tools, they can be steered in that direction. Compliance still has to be demonstrated by static analysis and review, not assumed. We show how to feed analysis results back so the agent fixes deviations itself.

Is the workshop only for ECU software?

No. Teams working on infotainment, connected services, vehicle backends or test automation benefit as well; for them, cybersecurity, data protection and supplier contracts take more room than safety.

Workshop

Get in touch

Want to use coding agents in automotive projects without risking your next assessment?

Tell us roughly how many people will join and pick dates that work – we will get back to you within 24 hours with a proposal for content, format and schedule.

Prefer to talk first?

15 minutes, free and without obligation – with our AI Trainers and AI Code Engineers.

Book a call for free (opens in new tab) +49 761 519 751 50 (24/7)

Request a training / workshop

Workshop: Agentic Coding in Automotive Software: ASPICE & ISO 26262 Workshop · 1 day

Dates that work for you (optional)

Suggestions about 2 weeks ahead – pick as many as you like

Reply usually within 24 hours · jocapps is ISO 27001 certified · Privacy