On 10 October 2026, we decided to pause active development of Project La Vinci's current parser-and-conversion path.
The Origin and the 1,000-File Test
La Vinci began as an attempt to create a reusable engineering-file layer around a normalized intermediate representation. We built a DWG ingestion pipeline, downstream rendering and conversion infrastructure, and tested the system across a 1,000-file corpus.
That work taught us something important: successful execution is not the same thing as source fidelity. Manual inspection exposed output that our automated validation had not adequately measured.
A pipeline completing without error codes does not prove geometric or semantic source fidelity. When manual inspection reveals discrepancy between source intent and rendered representation, shipping developer APIs against that foundation violates our engineering standards.
Customer Evidence and Market Pull
More importantly, our customer conversations have not repeatedly validated a new general-purpose CAD conversion engine as the infrastructure problem companies most need from Rine.
The strongest conversations are pointing toward a different boundary: engineering information inside drawings, specifications, attachments, and technical documents needing to become structured data inside another software workflow.
So we're stopping rather than allowing sunk engineering effort to become product strategy.
Preserved Assets and Next Direction
The La Vinci codebase, corpus, IR work, and testing infrastructure are being preserved. They remain useful technical assets. But further product development will now follow demonstrated customer requirements rather than expanding the stack for its own sake.
Rine's underlying thesis remains unchanged: software companies building in engineering-heavy industries repeatedly encounter difficult infrastructure around engineering information. Our job is to discover which parts are recurring, separable, and important enough for those companies to rely on Rine to solve.
Working with engineering documents or drawing sets?
Tell us what information your product needs to extract and what makes the workflow difficult today.