How it works
From commits to a substantiated claim, in about an hour.
No timesheets to reconstruct, no blank-page R&D report to write. We start from the work you already did and the trail it left in GitHub.
-
Step 1
Connect GitHub
Point us at one or more repositories. We support multiple repos for a single company.
-
Step 2
We ingest the work
We pull your pull requests (or your commit history, for repos that do not use them), the contributors behind them, and the timestamps. This is the raw evidence your claim is built from.
-
Step 3
We do a first pass
We automatically set aside the obviously routine work, like dependency bumps, lockfile changes, and docs-only edits, so the interview focuses only on what might be real R&D.
-
Step 4
Guided interview
We ask your engineers specific, grounded questions about the work that is left. Not "describe your R&D", but things like: "You reworked this module three times over six weeks with two reverts. What did you not know going in?" Your answers map each genuine experiment against the statutory four-gate test.
-
Step 5
Your claim package
You get a reviewable package: the confirmed experiments, their substantiation, the evidence each one links to, an estimate of hours and cost per contributor, and an estimated dollar offset. Your tax adviser reviews and lodges it.
What every experiment has to clear
Each activity we include is tested against the four things the law requires:
- Technical uncertainty
- The outcome was not known to a competent professional in advance.
- Systematic progression
- You worked from a hypothesis through experiment, observation and evaluation to a conclusion.
- New knowledge
- The purpose was to generate new technical knowledge, not just to confirm what was already known.
- Not excluded
- It is not one of the activities the law specifically excludes, like software purely for your own internal administration.
We are grounded in AusIndustry and ATO guidance, and we bias toward precision: a false positive (routine work claimed as R&D) is the dangerous error, so when in doubt we leave it out.