@Austin Tomko the classic ghl test button trap man... u are probably hitting test workflow instead of naturally triggering it from a test contact... when u manually test ghl doesnt pass the actual conversation ai payload or transcript to the next node... so the ai basically reads a blank slate and outputs not provided... u have to actually text the bot from a dummy contact so the system catches the real context... also double check u are explicitly mapping the output custom value from that summarize step directly into the email body... if u still hit a wall wiring those specific custom values just drop a dm... happy to show u the exact variable mapping u need :)

@Austin Tomko watched the video... ur core logic is fine but ur testing methodology is breaking the data payload... the test workflow button in ghl is strictly for linear logic... it forces the contact through the nodes instantly... but the conversation ai node requires live bi-directional data to actually generate the transcript... when u hit test the system bypasses the actual llm back and forth... so by the time it hits the summarize node the array is literally empty... hence the blank emails... u cant simulate ai latency with a test button man... u have to trigger it in the wild... drop ur actual cell number in a dummy contact... fill out the fb form live... let it text u... then physically reply back and forth 3 or 4 times so the ai actually builds the conversational memory and hits the qualification goal... once the goal is met and it naturally pushes to the notification node those variables will populate perfectly... stop testing ai pipelines linearly and ur build will work flawlessly :)