A multi-stack workflow, powered by a GPT bot, to feed an input video and receive a 99%-accurate, correctly formatted transcripts.
| Version | 1.0 |
| Automation type | Supervised |
| Availability | Fiverr |
| Services | GPT-4, Zapier, Imaginario AI |
| Pricing | $5 per 45 minutes |
| License | None (uses non-public features) |
background
we wanted to explore the usage of ai agents to accept and process orders for simple digital services, in this case video and audio transcription through Fiverr.
Fiverr integration with third-party platforms for communication is limited, so the customer service element of the automation is still mostly manual. However file ingestion, processing and delivery is possible through a workflow connecting various services together.
a different marketplace with extensible chat functionality would enable this tool to be 100% automated, however human supervision would still be necessary due to the usage of GPT-4, which can still produce hallucinatory results despite strict guardrails.
an alternate approach would be using a decision tree structure for customer interactions (limiting the ai functionality purely to transcription and tidy-up) however this would severely limit the scope of customer service.
Release notes
- 08/04/24 – version 0.9 released
- 17/04/24 – version 1.0 released
- Fixed Zapier formatting bug
- Added more guardrails to GPT-4 prompts
try it yourself
I will transcribe dialogue from any video or audio source
This transcription service is primarily powered by a series of automations, with human oversight. It will transcribe English-language dialogue with 99% accuracy.
Pricing from: $5 for 45 minutes
