We are testing AI meeting assistants against the same synthetic project-sync meeting. This page is the public log of what we can verify so far: setup friction, upload/import blockers, and the quality checks a human should run before sending AI-written follow-up notes.
We have not scored any meeting assistant for output quality yet because no tool has produced raw fixture output in a safe, reproducible no-spend run. That is useful evidence by itself: beginners should confirm whether a tool can accept their meeting format before trusting its action-item features.
Current best evidence: Notta reached a Free dashboard and upload modal with a neutral test persona, Otter login was recovered but import remained unclear, and tl;dv exposed only Google/Microsoft/SSO login methods in the scheduled test environment.
Our synthetic project-sync meeting includes defined decisions, owners, due dates, unresolved questions, and must-not-invent constraints. We score tools against an answer key, not against how polished a summary sounds.
Status: Free dashboard reached; upload blocked in scheduled browser
Setup: ~6 min recon
Score: Not scored
Neutral email signup worked and the Free dashboard/upload modal was visible, but the local fixture upload could not be completed through the scheduled file picker. The created account needs password reset/storage before reuse.
Status: Account recovered; import path unclear
Setup: ~15 min recon
Score: Not scored
Login to the existing neutral test account was verified, but the scheduled environment did not expose a reliable way to import the synthetic transcript or audio fixture.
Status: OAuth/SSO-only signup observed
Setup: ~3 min recon
Score: Not scored
The visible signup options were Google, Microsoft, and SSO only. We did not use personal OAuth, create an account, connect a calendar, or upload fixture data.
Status: Public-doc availability checked
Setup: Pending
Score: Not scored
Free upload appears promising enough for a later recon, but official export/download limits look Pro-gated, so it is a partial candidate rather than a recommendation.
Status: Blocked for uploaded fixture
Setup: N/A
Score: Not scored
Official docs point to live meeting capture rather than prerecorded upload, which does not fit this synthetic uploaded-fixture test without a safe live-meeting setup.
Status: Blocked on Free for uploaded fixture
Setup: N/A
Score: Not scored
The Free path requires live assistant/calendar-style testing; uploaded audio/video needs paid credits or a paid plan, so no no-spend benchmark output exists yet.
Status: Blocked for uploaded fixture
Setup: N/A
Score: Not scored
The product is designed around a live desktop meeting workflow, not prerecorded upload/import, so it is not a fit for this first uploaded-fixture pass.
A safe follow-up plan separates decisions, action items, unresolved questions, and risks. Before sending it, compare every owner, due date, price, launch date, approval, and policy claim against the transcript or your own notes.
The failure mode to watch for is false certainty: an AI note taker may phrase “we should check this” as if the team agreed to do it, or drop an important caveat because it makes the summary less tidy.
No affiliate links are used on this page. No payment method, personal account, OAuth login, calendar connection, real meeting data, or paid trial was used for these runs. Scores stay blank until raw fixture outputs and export evidence exist.
Last updated: May 29, 2026. Next evidence goal: recover/store the Notta test-account credential or use another safe upload path, then capture raw transcript/summary/action-item output before assigning any quality score.