AI assistant for café volunteers
Creation of an intelligent Telegram agent that automates volunteer support for a non-profit café, allowing the team to focus on hospitality rather than searching for operational answers.
Context
D.Café is a non-profit project where all profits go to charity. The team consists of volunteers who work in shifts and have varying levels of training. The core philosophy of the café is "radical hospitality." Before the implementation of the solution, volunteers spent time searching for answers regarding operational processes, which distracted them from interacting with guests.
Problem
Lack of quick access to a knowledge base: cash register procedures, equipment usage, menu standards, and communication scripts.
The need to constantly distract experienced team members for basic instructions.
Loss of focus on guests due to technical and organizational questions, especially during the onboarding of new volunteers.
Goal: to provide volunteers with instant access to all operational information through a familiar tool (Telegram) without the need for additional training.
Solution
A conversational AI agent has been developed that acts as a digital administrator. The system is based on a structured knowledge base containing a complete description of the café's processes. The agent uses semantic search, allowing it to understand the context of questions rather than just keywords.
Key Features:
RAG Architecture: responses are based solely on the uploaded café instructions, ensuring their accuracy.
Tone of Voice: setting a warm and welcoming communication style that aligns with the values of D.Café.
Zero entry threshold: using Telegram as the main interface, which does not require the installation of new applications.
Accuracy control: the agent is configured to report the absence of information instead of providing unverified data.
Implementation Process
Audit and structuring of knowledge: gathering instructions for opening/closing shifts, kitchen rules, cash procedures, and guest interaction scripts.
Building a knowledge base: organizing information into a vector database optimized for precise contextual search.
Agent architecture in n8n: using OpenAI Chat Model with Simple Memory integration to maintain conversation context and OpenAI Embeddings for semantic search.
Tone configuration: programming a specific style of responses (using phrases like "happy to help," "have a blessed day," etc.).
Testing and launch: deploying the bot, conducting briefings for volunteers, and collecting feedback directly during shifts.
Results
Automatic processing of 10-20 requests per shift that previously required administrator involvement.
Average response time of up to 60 seconds, allowing volunteers to quickly return to working with guests.
Minimal operating costs for the system due to optimized API requests.
Accelerated onboarding: new volunteers receive real-time support.
High level of trust in the system due to the accuracy of responses and adherence to corporate communication ethics.
#n8n #AI_Assistant #RAG #KnowledgeBase #OpenAI #Telegram_Automation #NonProfit_Tech #CustomerService_AI #SemanticSearch #VectorDatabase #VolunteerManagement
Context
D.Café is a non-profit project where all profits go to charity. The team consists of volunteers who work in shifts and have varying levels of training. The core philosophy of the café is "radical hospitality." Before the implementation of the solution, volunteers spent time searching for answers regarding operational processes, which distracted them from interacting with guests.
Problem
Lack of quick access to a knowledge base: cash register procedures, equipment usage, menu standards, and communication scripts.
The need to constantly distract experienced team members for basic instructions.
Loss of focus on guests due to technical and organizational questions, especially during the onboarding of new volunteers.
Goal: to provide volunteers with instant access to all operational information through a familiar tool (Telegram) without the need for additional training.
Solution
A conversational AI agent has been developed that acts as a digital administrator. The system is based on a structured knowledge base containing a complete description of the café's processes. The agent uses semantic search, allowing it to understand the context of questions rather than just keywords.
Key Features:
RAG Architecture: responses are based solely on the uploaded café instructions, ensuring their accuracy.
Tone of Voice: setting a warm and welcoming communication style that aligns with the values of D.Café.
Zero entry threshold: using Telegram as the main interface, which does not require the installation of new applications.
Accuracy control: the agent is configured to report the absence of information instead of providing unverified data.
Implementation Process
Audit and structuring of knowledge: gathering instructions for opening/closing shifts, kitchen rules, cash procedures, and guest interaction scripts.
Building a knowledge base: organizing information into a vector database optimized for precise contextual search.
Agent architecture in n8n: using OpenAI Chat Model with Simple Memory integration to maintain conversation context and OpenAI Embeddings for semantic search.
Tone configuration: programming a specific style of responses (using phrases like "happy to help," "have a blessed day," etc.).
Testing and launch: deploying the bot, conducting briefings for volunteers, and collecting feedback directly during shifts.
Results
Automatic processing of 10-20 requests per shift that previously required administrator involvement.
Average response time of up to 60 seconds, allowing volunteers to quickly return to working with guests.
Minimal operating costs for the system due to optimized API requests.
Accelerated onboarding: new volunteers receive real-time support.
High level of trust in the system due to the accuracy of responses and adherence to corporate communication ethics.
#n8n #AI_Assistant #RAG #KnowledgeBase #OpenAI #Telegram_Automation #NonProfit_Tech #CustomerService_AI #SemanticSearch #VectorDatabase #VolunteerManagement
Lvov 
Available for hire