The dashboard now sees the call happening, waits visibly for the transcript, and files it in the lead's history · 2026-07-06
The dashboard now sees the call happening, waits visibly for the transcript, and files it in the lead's history
Master summary — the gist in 30 seconds
TL;DRYour calls were invisible to the pipeline. Now a booked call lights the lead's card purple ('waiting for transcript'), then drops the transcript into the card and the lead's history automatically.
Booked call time + Fireflies recording → a purple 'waiting' card → the transcript on the card and in the touchpoint history, plus yellow reminder rows and a working confirmation-email edit.
Why this mattersBefore this, because you record with bot-free Fireflies (not our in-tab script), the system had no idea a call ever happened — so nothing showed on the board. Now the board reacts to the booked time itself, so it works with zero setup, and gets even sharper (exact hang-up) once you add one small script.
flowchart LR
A[Booked call time arrives] --> B[Card turns purple\nwaiting for transcript]
B --> C[Fireflies finishes]
C --> D[Transcript on card\n+ in lead history]
A -.optional script.-> E[Exact hang-up\nwithin 30s]
E --> B
1 · Why the call was invisible
TL;DRThe board only knew about a call if our in-tab Meet script pinged it — which you don't run.
Live Meet ping → 'call started' stamp → card lifecycle. No ping → no stamp → blank card.
Why it mattersThe whole feature hung on a single signal you weren't sending. The fix stops depending on it: the booked call time is a signal we always have, so the card can react even with no script running.
TL;DRAfter the booked end time, the card goes purple with an '⏳ waiting for transcript' tag, and refreshes itself every ~12 seconds without reloading.
Now past the call's end + no transcript yet → purple overlay + tag. Gives up quietly after 45 min if nothing ever arrives.
Why it mattersIt's the visible bridge between hanging up and the transcript landing — you can glance at the board and see 'a call just happened, we're waiting on the write-up' instead of a dead card. The self-refresh means you never hit reload (which would wipe an edit in progress).
stateDiagram-v2
[*] --> InProgress: call time now
InProgress --> Waiting: end time passed
Waiting --> TranscriptReady: Fireflies delivers
Waiting --> [*]: 45 min, nothing came
3 · Transcript + reminders in the history
TL;DRThe transcript now files itself as a history touchpoint, and reminder emails show as yellow rows with each one's time, subject and text.
Transcript arrives → 'call completed' history row. Booking made → yellow reminder rows (when + subject + body).
Why it mattersReminders were always being scheduled — they were just logged invisibly (grey, no detail) and the yellow styling was literally dead code. Now the history reads like the real conversation: what was sent, when, and the call write-up, all in one place per lead.
flowchart LR
BK[Booking] --> R[Yellow reminder rows\ntime + subject + text]
FF[Fireflies transcript] --> H[Call-completed row\n+ transcript on card]
R --> HIST[Lead touchpoint history]
H --> HIST
4 · The one thing that needs you
TL;DRInstall a small Google Meet script once to get exact within-30-seconds-of-hang-up detection; until then, purple appears at the scheduled end time.
Paste + authorize the script (steps in TRANSCRIPTION_WIRING.md) → the pipeline knows the moment you actually hang up.
Why it mattersThe scheduled-time version needs zero setup and already works. The script upgrades it from 'roughly at the booked end' to 'the second you leave the call' — the exact behaviour you asked for. A real 2-person test call is the only thing I can't verify for you automatically.
flowchart TD
Now[Today: purple at booked end time\nno setup needed] --> Better[Add Meet script once]
Better --> Exact[Exact within-30s hang-up\n+ real 'in progress' state]
Exact --> Test[You: one real test call to confirm]