Bonsai Agent automates TARA, HARA, FMEA and more for safety-critical systems. Standards-aligned results in minutes, not months.
Expert agents available today
Identify hazards in the braking system, approve the agent’s proposed safety goals, and export them to SystemWeaver with their ASIL ratings.
Ask the agent anything about your project
Connect your ALM, codebase, PLM and engineering apps. Bonsai reasons across all of them, from requirement to code to field fault.
Polarion, Codebeamer, SystemWeaver
Git repositories, C/C++ sources, static analysis
BOMs, variants, change orders, field data
Issue trackers, test benches, document stores

One traceable context
Derive and decompose requirements, traced back to safety goals and security claims.
Keep system and software architecture consistent with your safety and security concepts.
Review code against MISRA and link every change to the requirement it implements.
Assemble assessment-ready work products and evidence as the project evolves.
Follow a fault from the field back to its root cause across every connected system.
Deep embedded knowledge of each standard, ready without setup. Upload your own guidelines and every agent follows them.
ISO 26262
HARA, ASIL, safety concepts, FMEA, FTA and hardware metrics.
ISO 21434 · UN R155 · AIS-189
TARA, CAL and attack trees, plus the CSMS type-approval evidence UN R155 and AIS-189 require.
UN R156 · AIS-190
Software update processes, RXSWIN, OTA rollback and type-approval impact of changes.
Automotive SPICE
Process areas, capability levels, base practices and work products.
ISO 21448
ODD, triggering conditions, scenarios and ADAS/AD validation.
C:2012/2023, C++:2023
Rules, rationale, code review and deviation guidance.
Classic Platform
SWC design, BSW, RTE, OS configuration and ARXML.
Adaptive Platform
ara:: APIs, SOME/IP, SOA and execution management.
In development
DO-178C, ARP4754A, ARP4761 and DO-326A airworthiness security.
Bonsai Agent picks up your formatting, terminology and domain conventions as you work, and applies them to every future conversation.
Review, edit or clear what it has learned at any time.
Ask the agent anything about your project
Nothing reaches your ALM without an independent review and your approval.
Fully air-gapped with local LLMs. Your data never leaves your infrastructure.
State-of-the-art cloud models, with all project data stored on your infrastructure.