Built for investigation · Early access
EmbedVera / AI CAN Debugger
From CAN issue
to clearer evidence.
Investigate CAN issues from logs, DBC files and function specifications with traceable engineering evidence.
Local-first. Evidence-driven. Engineer-led.
Customer issue
“The requested function responds inconsistently.”
- FACT
- A request is visible in the selected log window.
- INFERENCE
- The response path needs closer investigation.
- UNKNOWN
- Internal execution and final root cause.
- NEXT_INVESTIGATION
- Obtain the internal observation needed to test this direction.
Schematic only · No customer data or measured result
01 / Start with context
Your engineering context.
One focused investigation.
Start with the issue. Bring only the evidence needed to understand it.
Customer issue
Describe the symptom, operating conditions and the question you need to answer.
Engineering questionASC log
Observe the relevant signal transitions, timing and request/response windows.
.asc · Text bus logDBC
Turn raw message payloads into named signals and values for your project.
.dbc · Signal databaseFunction specification
Find relevant requirements and assess supported conditions against their source.
.docx · When available02 / Follow the evidence
Know what happened.
Know what to investigate next.
Relevant signals
Focus on candidate signals and time windows connected to the customer issue.
Expected vs observed
Keep measured behavior separate from requirement-backed expectations. Without an expectation source, abnormality remains unassessed.
Evidence-backed candidates
Keep facts, inferences and unknowns distinct. Check the observations and source references behind a candidate explanation.
A useful next step
Move the investigation boundary with a concrete request for missing evidence, rather than a forced root-cause claim.
03 / Built around engineering judgment
Keep your evidence.
Keep your judgment.
Read the data boundaries Local-first by design
Raw engineering artifacts stay local by default. Live AI is an explicit choice and receives selected relevant context.
Traceable to the source
Reports retain evidence references and project-relative source locations, confidence, unknowns and limitations.
Conservative about causality
Correlation can support a direction. It cannot, by itself, prove a software defect or an internal execution path.
Field notes
A better starting point.
Questions, answered
Frequently asked questions
What does EmbedVera do?
EmbedVera is an AI-assisted CAN issue investigation tool for embedded and automotive engineers. It connects a customer issue, ASC logs, DBC files and available function specifications to traceable observations, candidate explanations and a next investigation step.
Which inputs does the current Beta support?
The current engine works with tested ASC text logs, DBC signal databases, DOCX text function specifications and explicit engineer context. Other log formats, source-code analysis, Simulink and Stateflow analysis are not currently supported.
Does EmbedVera automatically find the root cause?
No. It helps narrow an investigation and preserve supporting evidence. Timing correlation and a missing observed response do not prove internal software causality. Final engineering judgment stays with the engineer.
Are raw engineering files uploaded?
Raw engineering artifacts are local-first, and large logs are not sent wholesale to a model. If live AI is explicitly enabled, selected relevant context may be sent to the configured model provider. This website does not accept engineering file uploads.
Early access Beta
Bring a question.
Leave with a direction.
Help shape an investigation tool for embedded and automotive engineers.