Pilar v0.1
Safety
How PILAR tries to be trustworthy — and the limits you must keep in mind.
Checked by design
Two independent AI engineers solve each problem with different methods. A comparator confirms they agree before anything is shown — a disagreement halts the run — and a controller then decides whether it is safe to present the result and flags caveats a reviewing engineer should know about.
Always preliminary
Every output is AI-generated and preliminary. It must be reviewed by a qualified engineer before use in real projects.
What PILAR does not do
PILAR is not a substitute for a qualified engineer's judgment. It does not produce verified design documentation, and it does not warrant that results are correct, complete or conformant with the applicable codes. Use is at your own risk.
Data handling
Internal monitoring tools surface sanitized, read-only summaries only. Human review remains final.