Software for grant, SR&ED and tax credit consultants
Every claim is an evidence package assembled under a deadline from clients who answer slowly. The consulting is the easy part. The operational problem is collecting proof from busy engineers before the filing window closes.
Capital and credit. A file opens, documents get collected, a third party decides, money moves, and every step has to be auditable. We have run this machine in production.
Sub-niches covered: SR&ED practices, CDAE and provincial tax credits, IRAP and innovation funding advisors, Grant writing firms, Export and market development funding, Film and media tax credits.
You likely have this problem if:
Every program gets the same generic document request.
Technical narratives are written from interviews nobody recorded.
Filing deadlines across the book sit in one person's calendar.
Time allocation arrives in whatever format the client's bookkeeper likes.
You cannot see which claims are short of evidence with a week to go.
What breaks operationally:
Evidence requirements differ per program but the collection process is the same generic email every time.
Technical narratives are drafted from interviews that were never recorded or structured.
Filing deadlines across a book of clients are tracked in one person's calendar.
Time and payroll allocation data arrives in whatever format the client's bookkeeper prefers.
What we build:
Per-program claim workspace. A workspace per claim with the evidence requirements of that specific program, not a generic folder.
Client evidence collection. Structured upload with program-specific checklists and follow-up that escalates as the deadline approaches.
Deadline board across the book. Every client's filing window in one view, ranked by how much evidence is still outstanding.
Narrative capture. Structured technical interviews captured against the claim, so drafting starts from a record rather than a memory.
Can AI draft SR&ED technical narratives?
It can draft from structured input and it is genuinely useful there, but the input is the constraint. A narrative generated from a thin interview is confidently wrong in ways a reviewer notices. The leverage is in capturing better raw material, and then drafting is fast either way.
How do you handle clients with no time tracking?
You collect what exists — payroll, project records, commit history, calendars — and make the allocation assumptions explicit on the claim rather than invisible. Reviewers respond much better to a stated methodology than to a precise-looking number with no basis.
Is a per-program checklist worth building for small programs?
It is if you file the program more than a handful of times a year. The checklist is what stops a junior consultant from collecting the wrong evidence, and that mistake costs more than the setup.
What about multi-year claims and amendments?
They need to be first-class in the model, not a copy of last year's folder. Amendments in particular fail when the system cannot show what changed and why.
Last reviewed 22 August 2026