SR&ED for Software & SaaS

Yes. SaaS companies often have strong claims when their teams solve technical problems that standard tools and practices can't, such as scalability, performance or data processing challenges.

Yes. Sprint tickets, commits, pull request discussions and test results are good evidence. The key is making sure the technical uncertainty and what you tried are recorded, not just the feature that shipped.

It can be, when your team had to go beyond standard practice to reach the result. Training a standard model on your own data usually isn't enough; developing new approaches because established ones didn't work can be.

Not all software development qualifies — routine coding, bug fixes, and standard application development generally do not. What qualifies is work that attempts to overcome a technological uncertainty that cannot be resolved through standard practice. This includes novel algorithm design, building new language runtimes, pushing the boundaries of ML model architectures, or optimizing for performance targets not achievable with existing libraries.

Prior to Bill C-15 (2026), capital costs like GPU hardware were not eligible. With the restoration of capital expenditure eligibility for assets acquired after December 15, 2024, compute infrastructure used directly in SR&ED activities can now be included in your claim. Ongoing cloud subscription costs are a separate category — we can assess the specific treatment for your setup during a free consultation.

Agile development is absolutely compatible with SR&ED claims. The key is documenting technical hypotheses, failed approaches, architectural decisions, and the uncertainty you were navigating at each sprint. Version history, pull request comments, technical design documents, and retrospective notes all serve as contemporaneous documentation. SoftInventors helps you set up a lightweight documentation process that fits your workflow without slowing down your team.