Purpose
The primary purpose is the implementation and delivery of the inbound enquiry and booking
chatbot agent for Barbeque nation, which would be answering FAQ Questions, Noting down
new bookings, and updating/cancellation of bookings for the properties catered by Barbeque
Nation in Delhi, Bangalore.
System Breakdown
ition Objectives
te and a state prompt represents the context or instruction of
onversational AI system at a given point during an - Maintain Instructional Integrity: Accurately capture
ction. and store the current context to ensure the system
nversational flow would have multiple states, and at one turn understands the conversation's status and the action
onversation only one state can be active. steps to take for resolution.
- Ensure only one state is active at any given turn to
prevent ambiguity in the conversation flow
te transition prompt is a predefined instruction or template - Correctly evaluate user input and contextual data to
by the conversational AI to determine how to move from one identify the appropriate next state, ensuring smooth
to another based on the current state and user input. It is a progression of the conversation
r logic component, often implemented as a conditional - Reliably determine when the current state’s intended
ment that evaluates the user’s input and contextual data to task is fulfilled to trigger the transition to the next
t the appropriate next state. state.
the intended task of the state is complete, moving to the
state is the task of the state transition prompt.
nowledge base is a structured of information that the
ersational AI system draws upon to generate accurate and - Deliver accurate and contextually appropriate data to
xtually relevant responses. It contains domain-specific data, support the system’s response generation or query
rules, or documents (e.g., FAQs, product manuals, or resolution.
ogies) and may be implemented as a database, vector store,
aph structure. The system queries the knowledge base using
iques like information retrieval or semantic search to
ent its responses or resolve user queries.
call analysis refers to the process of evaluating and - Giving clarity to the business on the agent’s
cting insights from a completed conversation (or "call") performance and if any resolution pendings are
een the user and the conversational AI system. This involves supposed to be completed from their team’s end.
zing metrics such as user satisfaction, intent fulfillment,
nse accuracy, conversation flow, and error rates.
niques may include log parsing, sentiment analysis, topic
ling, or performance benchmarking. The analysis is used to
fy areas for improvement in the system’s dialog
gement, knowledge base, or state transition logic, often
ng into model retraining or system optimization.
rse engineer the API endpoints of the platform you are
ng the agent on, using the “Network” tab to create a chatbot
hare the Chatbot Link.
Functional Requirement
● Fork the templates and the methods shared in
https://github.com/AryamannNingombam/agentops-template-repo
to complete the API endpoints for the Knowledge Base, Chatbot and Post-Call
Configuration.
● To build the steps of the conversation flow, utilise the:
○ Conversational Flow Diagram.
○ State Template Diagram
to drive the conversational flow.
● To drive the conversation flow in the right direction, write down the state transition
prompts which would lead the conversation in the right direction.
● For each property, convert the raw knowledge base shared by the business into a
structured knowledge base tool which would be used by the agent to answer enquiries
based on the property they are enquiring regarding.
● Host the API endpoints for the knowledge base and the phone number for
conversational flow.
● BONUS Functional Requirements:
○ Post-Call Analysis which would share information regarding the interaction
status.
○ Chatbot creation where you will reverse engineer the API endpoints of
beta.retellai.com to create a chatbot with your own frontend UI and host it for
demonstration.
Submission Method and Guidelines
- Create a Fork of: https://github.com/AryamannNingombam/agentops-template-repo
- Add the link for the
- Knowledge Base API Endpoints
- Post-Call Analysis Excel Sheet
- Chatbot Link
- Agent Linked Phone Number
in your project ReadME description.
- Submit this form: https://forms.gle/7qSTquHbBaKzy9MJ8
- A short technical documentation which tells the complete story of your system. Include:
- Detailed architecture diagrams showing how all components interact
- API documentation explaining all endpoints and their purposes.
Resources
- State Prompts & Knowledge Base : AgentOps - Assignment Resources
- Platform for State Machine: https://beta.retellai.com/
