Development of the Binotel call analysis system based on Gemini AI (MVP with editable criteria)

Budget: 14700 UAH Deadline: 7 days

Inch.

I carefully read your technical specifications, and the task is clear to me. I propose to create a simple and reliable quality control system for calls without unnecessary complexity — so that it is easy to maintain and scale further.

How it will work in practice

The logic is exactly as you described:

the call ended → the system automatically retrieved the recording → AI analyzed it → the result appeared in Google Sheets.

We receive audio after the call through the Binotel API. The recording is then sent to Gemini 1.5 Flash, which performs the analysis based on the specified criteria, and all results are automatically recorded in Google Sheets.

Main focus — flexible criteria

There will be a separate tab in the sheet called "Settings," where you can change:

* the wording of the criteria,
* descriptions of the ratings,
* analysis focuses.

The script reads these settings before each analysis and forms a prompt for the AI.
That is, you change the text in the sheet — and the next call is analyzed according to the new rules, without rewriting the code.

What exactly the system evaluates

* Overall customer orientation (1–10)
* Whether an order occurred (Yes / No)
* If not — a brief reason
* The manager's politeness
* Handling objections and approach to the client
* A brief comment from the AI on the call

Each call is a separate row in the sheet, everything is transparent and convenient for viewing.

Weekly summary

Once a week, the system:

* analyzes all calls for the period,
* generates a short understandable text report: what improved, where there are problems, which recurring moments,
* either records this report in a separate tab,
* or sends it via Telegram.

Architecture

Everything is as simple as possible:

* Python script
* Google Sheets API
* Binotel API
* Gemini 1.5 Flash
* no databases and complex services

Terms and budget

Budget: $300 for the project
Implementation time: 5–7 working days

The cost includes a fully working system, table setup, report logic, and a brief instruction on how to work with it.

If the solution proves beneficial, we can move towards deeper analytics and automation — the foundation for this will already be laid. Expand

Budget: 7000 UAH Deadline: 1 day

Inch.
Hello. This is a ready solution, it needs to be customized for you.

Budget: 7000 UAH Deadline: 7 days

Inch. I can implement the system so that the criteria are pulled directly from the "Settings" tab in Google Sheets before each analysis, without manual edits to the code. After the call ends, the script will automatically retrieve the recording, send it to Gemini for analysis, save the structured result in the table, and create a history for future reports. The weekly report can also be automated on a schedule with the generation of a summary text and sending it to Telegram or recording it in a separate tab.

Budget: 7000 UAH Deadline: 4 days

Inch.
Hello!
I have experience integrating the Binotel API and working with Gemini 1.5 Flash (this is the perfect choice for audio because it is cheap and understands context).

My solution:

1. We catch the "Call completed" event from Binotel.
2. We send the audio file directly to Gemini (without unnecessary transcription services). This allows the AI to hear not only the text but also the tone of the conversation.
3. The script reads the evaluation criteria directly from your Google Sheet each time. If you change the description in the cell — the next call is evaluated according to the new criteria.
4. Once a week, the script generates a report (trends, issues) and sends it to you on Telegram.
Everything will work on a simple cloud script, without complex databases, as you requested. Expand