SimPilot includes multiple handoff paths depending on what you need to share.
Chat exports
From a chat export action, you can export as:
- PDF (rendered conversation view)
- JSON (structured data)
- Markdown (human-readable text)
Use JSON for machine workflows and Markdown/PDF for human review.
Project exports
Project-level export bundles all chats into a ZIP archive. Each chat is included in multiple formats for portability.
Report generation
SimPilot produces professional simulation reports as rich document artifacts. Reports are auto-generated on successful workflow completion and can also be requested on demand for any completed simulation.
How reports are generated
Planning
The report planner analyzes the conversation context and simulation data to create a structured report plan -- selecting sections, required figures, and whether web research is needed for validation context. Skill-based domain context (solver documentation, workflow references) is loaded automatically based on the simulation software detected in the case.
Data gathering
The data gatherer resolves the target simulation job, collects residual histories, mesh quality metrics, solver parameters, and quantities of interest. For studies with multiple experiments, it cross-references to find the best representative job.
Figure generation
Figures are generated automatically from the gathered data -- convergence plots, quantity charts, and field visualization screenshots.
Compilation
The report is compiled using the planned structure, gathered data, and generated figures into the selected output format.
Study-aware reports
When a report is generated from a chat that contains a
conduct study session, the report pipeline automatically cross-references study experiments and their simulation jobs. It selects the most representative job -- preferring the study's best experiment result or the explicitly linked simulation job -- so the report reflects the study's conclusions rather than an arbitrary run.
Report sections
| Section | Contents |
|---|
| Executive summary | Key findings and simulation outcome |
| Setup | Solver, boundary conditions, physical properties |
| Mesh quality | Cell count, quality metrics (non-orthogonality, skewness, aspect ratio) |
| Solver settings | Discretization schemes, linear solver configuration, relaxation factors |
| Convergence | Final residual levels, iteration count, convergence history |
| Results | Extracted quantities of interest (drag, lift, pressure drop, etc.) |
| Provenance | Knowledge base sources that influenced the configuration |
- PDF (primary) — publication-quality PDFs compiled with LaTeX using a single deterministic publication template
- JSON — native rendering as a rich document artifact in chat with charts, metric cards, and status badges
- HTML — standalone page suitable for sharing (fallback when LaTeX compilation is unavailable)
- Markdown — plain text format
Reports support optional branding (company name and logo URL). The report generator includes an optional web research step to gather validation context and automatically generates figures from simulation data.
Auto-generation
When a simulation workflow completes successfully, a report is automatically generated and attached to the conversation. The agent keeps its response concise and does not restate information already present in the report.
Typical handoff patterns
- Design review: share link + PDF export
- Archival: project ZIP export
- Automation: JSON export into downstream systems
- Client delivery: report artifact + selected visuals