Taxnova.ai automates R&D tax credit claims by connecting read-only to engineering and financial systems to identify, describe, and size qualifying R&D activity. The platform generates evidence-linked narratives and staff time allocations, which advisors review and submit, reducing claim preparation from months to as little as two weeks. It currently supports claims in the US, UK, Canada, France, and Spain.
Funding
Funding not disclosed
Founders
Product
Problem
R&D tax claims typically take three to six months of coordination, requiring engineers to spend weeks on surveys and interviews. Evidence often depends on what people remember rather than actual system data, leading to misinterpretation by consultants, painful restatements when accruals differ from actuals, and heightened audit risk.
Solution
Taxnova.ai automates the R&D claim process by connecting with read-only access to engineering and financial systems, identifying and sizing qualifying R&D activity across teams and countries. The platform automatically generates narratives linked to evidence and staff R&D time allocations, which are then reviewed, finalized, and submitted by the advisor of the user's choice. Setup takes approximately 30 minutes, and claims can be completed in as little as two weeks for larger organizations. The system is designed to complement existing advisors rather than replace them, improving claim quality while reducing manual effort.
Target Audience
Finance and engineering leaders at modern technology companies, whether they file claims themselves, work with an R&D tax advisor, or are new to tax claims.
Features
- Read-only integration with engineering systems and financial ledgers to automatically identify and size eligible R&D projects
- Automatic generation of claim narratives linked to underlying evidence and staff R&D time allocations
- Country-specific alignment for the US, UK, Canada, France, and Spain, with 10 more countries planned by end of 2027
- SOC 2 Type 2 certified, with encryption in transit and at rest, annual third-party penetration testing, and data never used to train AI models
- Scoped, revocable access tokens with no source code access and no write-back capabilities
- Ledger scanning to surface potentially eligible cloud and software costs, with user control over final R&D percentage assignment