Groundwork is the memory layer for seed-stage product teams: every decision logged with its reasoning, every piece of research connected to what it changed.
If only “why was this decided?”
ever had an answer.
There is no single switch,
but there is a system that remembers.
Groundwork keeps the reasoning, the evidence,
and the link between them.
A decision made on a call, gone by the next sprint. What survives is the what. Never the why, or what you almost did instead.
Five sharp interviews become a doc nobody opens twice. The next debate starts from opinion again, as if the evidence never existed.
Nobody can say which piece of evidence changed which call. So the team relearns the same lessons. At full price, every time.
Say what happened in one field: Groundwork drafts the entry, you confirm and edit. Reasoning, alternatives considered, who was in the room. When your board asks what shipped and why, the update writes itself.
The real problem isn’t running the session.
For most first-time researchers, the question itself is the hard part. A hypothesis phrased as a leading question. A usability task written to flatter the design instead of testing it. Get the question wrong and you walk away confident in the wrong thing, and build accordingly. Groundwork writes the sharp version instead: a falsifiable question, not a leading one, sequenced the way a trained researcher would ask it.
A chatbot writes your research plan once and forgets it ever existed. Groundwork remembers what you learned, who said it, and which decision it changed. And the memory compounds every week you use it.
Decisions logged as they happen. Research saved the moment it’s made.
Evidence attaches to the decisions it informed. Linked, not filed.
The memory grows weekly. New teammates onboard from the log.
Seed to Series A product teams, roughly 10 to 40 people, where decisions and research start slipping through the cracks as the team grows. If you’re a founder or PM who keeps answering “why did we do it this way?” from memory, it’s for you.
We’re inviting teams off the waitlist in small batches. The questionnaire when you join helps us find the closest fit first. Those teams hear from us within days, not months.
Free during early access. After that: per-seat pricing for the Decision Log and a flat monthly price for the Research Assistant. Early-access teams keep a discount.
In an encrypted Postgres database, isolated per workspace. Your decisions and research are never used to train models, and meeting transcripts are only pulled when you explicitly connect and select them.
No. It links to them. Groundwork is the index of what you decided and learned; your PRDs and docs stay where they are, attached to the decisions they belong to.