Skip to main content
Evidence management

Turn scattered activity into linked SR&ED evidence

Most teams already create useful evidence.

Evidence workspace / Northwind Robotics
Illustrative

GitHub repository

Read-only activity mapped to the project.

Linked

Project records

Files, notes and links stay with the work they support.

Classified

Review gaps

Missing support stays visible before drafting.

Flagged

The problem is that commits, tickets, design notes, test results and time records live in different systems and rarely use the language of Form T661.

SREDlog brings approved records into one project view, proposes classifications and keeps the human review attached to each link.

Bring in records from where the work happened

Connect GitHub for a read-only sync of commit and repository activity, upload documents and spreadsheets, or reference a source by link. Once you map a repository to a project, later synced activity arrives in the right claim.

Treat every classification as a proposal

  • AI reads each item and proposes its document type, the technical work it shows, and its relevance to SR&ED.
  • Tables and hours are pulled out where they exist.
  • Every classification is a proposal a person confirms; nothing is assumed.

Keep the source attached to the draft

Evidence is tied to the project it supports, then to the narratives and figures. The drafted T661 can cite the evidence behind a sentence, and the binder traces each figure to its source.

Frequently asked questions

The client installs the GitHub App, selects repositories and controls the grant. The sync is read-only, uses short-lived installation tokens and does not require personal access tokens.

Documents, dev logs, tickets, timesheets, images and PDFs (and CSV, XLSX, DOCX, TXT). You link each item to the project it supports.

No. It proposes classifications for technical work in your evidence; you confirm the classification, the project link and any extracted hours.

Evidence is isolated per firm, encrypted at rest and streamed only through authenticated routes. SREDlog does not pool client data across firms or use it to train models. OpenAI states that API data is not used for training unless the API account explicitly opts in. See security for provider-retention details.

Start with one project and its real records

Open a free trial, connect an approved source or upload documents, then review each project link.

7-day free trial · No credit card required