Promptly
A system for organizing and managing Telegram bots with AI. Consultations, sales, and service appointments 24/7 automated. Requires no programming or complex setup.
The system is under active development. There is no stable or demo version yet, as this is the second implementation with a changed stack.
The core idea of the project is to create and administer any number of bots in Telegram, without programming.
All settings are done exclusively through text, using prompts for AI.
The built-in semantic model and wrapper over Gemini allow for the creation of smart, context-aware AI assistants for business.
What a bot configured in Promptly can do:
- Provides 24/7 consultations using a specified tone and character.
- Understands about 200 languages; semantics are optimized for Kazakh, English, and Russian.
- Uses a trigger system for logging and extracting user data - name, age, city, etc. Any trigger can be configured separately for any task and business area.
- Uses a 'thinking' trigger to create an interface directly in Telegram - images and inline buttons.
- Can book a user for a service, event, or any specified time slot in the settings. Time slots can be configured regularly, cyclically, or as individual events on a specific day.
- Can offer products and create a user's shopping cart. Products can be categorized, with images, descriptions, and prices attached.
- The system has good analytics, data visualization, and a mini CRM that shows lead completeness based on configured logging triggers, to understand how ready a client is for a deal or purchase.
Technical nuances:
Theoretically, the number of bots created in the system is unlimited, but if SSL is not used on the server, all bots will be managed by a single dispatcher, which may affect performance.
Without encryption, it is not recommended to run more than 5 bots in the system.
No programming or node system, just simple text instructions and settings, all broken down into steps and translated into three languages.
The functionality described above is 100% working and implemented. I am not publishing the project completely because there are frontend issues after I fully transitioned development to Python + FastAPI.