Budget: 300 EUR Deadline: 10 days
Good day.
The task is clear. This is not just transcription, but the construction of a full-fledged quality control system for calls with a clear evaluation logic and stable data processing.
I work with similar solutions: AI analysis, event processing, integrations, building backend for such processes.
Here’s how I would approach the implementation:
1. Formation of the technical specification
— defining checklists (what exactly we are checking)
— evaluation structure (points, statuses, criteria)
— scenarios: norm / error / conflict
2. Transcription
— using Whisper or analogs
— support for Ukrainian / Russian / Czech
— preparing text for analysis
3. AI analysis
— checking the dialogue against the checklist
— determining whether key questions were asked
— identifying negativity / conflicts
— forming a structured result
4. Integrations
— obtaining recordings from Binotel / Bitrix24
— processing through Python (FastAPI)
— sending results to Telegram or Google Sheets
5. Architecture
— database (PostgreSQL)
— processing queue to avoid losing calls
— logging
— scalability options
I create such systems through a pipeline approach with clear processing states to avoid data loss and “partial results.”
What you will receive as a result:
— a working AI tool
— structured evaluation of calls
— automatic reports
— a foundation for further integration into CRM
Timeline:
7–14 days for MVP (transcription + basic analysis + reports)
Cost:
$400–900 depending on the depth of logic and integrations
I can start with a short audit and propose a specific architecture and a sample report before the development begins.