Mitra: AI FAQ Chatbot for Customer Self-Service Mitra is a website-embedded AI FAQ chatbot built for Nfilade Security Solutions / MyNetSec to help customers resolve common Tier-1 support queries through self-service.

The chatbot was implemented using Botpress Cloud Webchat and integrated into the company website with lightweight HTML and JavaScript. It gave customers a simple conversational interface for frequently asked questions, reducing repetitive manual support workload and helping resolve approximately 50% of Tier-1 support queries without human intervention.

Repository Summary Project type: Conversational AI chatbot integration Use case: Customer support automation Platform: Botpress Cloud Frontend integration: HTML, JavaScript Chatbot name: Mitra

Business outcome: Around 50% Tier-1 support query resolution through self-service

Problem Customer support teams often handle the same basic questions repeatedly. These questions are usually important, but they do not always require manual support. Typical examples include: • How do I contact support? • How do I access my account? • Where can I find service information? • How do I raise a support request? • What should I do if I face a login issue? Before Mitra, customers depended more heavily on static FAQ pages or manual support channels. This created unnecessary waiting time for users and repetitive work for the support team.

First-Principles View The real problem was not “build a chatbot.” The real problem was: Customers needed quick answers to repeated support questions, and the support team needed fewer low-complexity tickets. A full custom NLP system would have been excessive for this scope. A faster and more maintainable solution was to use a conversational AI platform, configure the support flow, and embed it directly into the customer-facing website. That is why Botpress was a practical fit.

Solution Mitra was built as a Botpress-powered FAQ assistant and embedded into the company website using the Botpress Webchat script. The chatbot acts as a first-response support layer. Users can ask questions in natural language, receive guided responses, and resolve common issues without submitting a ticket. For unclear or unsupported questions, the chatbot can guide the user toward human support or the appropriate escalation path.

Key Features • Website-embedded chatbot interface • Branded assistant named Mitra • Conversational FAQ support • Customer self-service for common Tier-1 queries • Botpress Cloud Webchat integration • Session-based chat experience • Lightweight HTML and JavaScript deployment • Scalable first-response layer for support automation

Tech Stack Area Technology Chatbot Platform Botpress Cloud Webchat Embed Botpress Webchat Frontend HTML, JavaScript UI Theme Prism Session Handling Browser session storage Deployment Context Website-based customer support Use Case FAQ automation and Tier-1 support deflection

System Flow Customer visits website | Botpress Webchat loads | Customer asks a support question | Mitra matches the query to a relevant FAQ or support flow | Chatbot returns a structured response | Customer resolves the issue or escalates to human support

Webchat Integration The chatbot was embedded using Botpress Cloud’s webchat script and configuration file.

<script src="https://cdn.botpress.cloud/webchat/v1/inject.js"></script> <script src="https://mediafiles.botpress.cloud/b53b849f-1eb2-493a-9746-fc818d576c5c/webchat/config.js" defer></script>

A simplified initialization example is shown below:

<script> window.botpressWebChat.init({ composerPlaceholder: "Chat with Mitra", botConversationDescription: "Customer support chatbot for FAQ self-service", botName: "Mitra", themeName: "prism", themeColor: "#2563eb", useSessionStorage: true, showBotInfoPage: true }); </script>

Clean HTML Embed Example

<title>Mitra FAQ Chatbot</title> <script src="https://cdn.botpress.cloud/webchat/v1/inject.js"></script> <script src="https://mediafiles.botpress.cloud/b53b849f-1eb2-493a-9746-fc818d576c5c/webchat/config.js" defer> </script> <script> window.addEventListener("load", function () { window.botpressWebChat.init({ composerPlaceholder: "Chat with Mitra", botConversationDescription: "Customer support chatbot for FAQ self-service", botId: "b53b849f-1eb2-493a-9746-fc818d576c5c", hostUrl: "https://cdn.botpress.cloud/webchat/v1", messagingUrl: "https://messaging.botpress.cloud", clientId: "b53b849f-1eb2-493a-9746-fc818d576c5c", webhookId: "0bb8fa12-c2b5-4dd9-923f-4fd1679c208b", lazySocket: true, themeName: "prism", botName: "Mitra", frontendVersion: "v1", useSessionStorage: true, showBotInfoPage: true, showPoweredBy: true, theme: "prism", themeColor: "#2563eb", allowedOrigins: [] }); }); </script>

Measuring Impact The chatbot’s value was measured through support deflection. Primary success metric: Tier-1 self-service resolution rate = resolved chatbot queries / total Tier-1 support queries Observed result: Approximately 50% of Tier-1 support queries were resolved through self-service This means the chatbot helped reduce repetitive support workload while giving customers faster access to basic help.

My Contribution My contribution included: • Configuring the Botpress chatbot experience • Embedding the chatbot into the customer-facing web interface • Customizing the assistant identity as Mitra • Setting up webchat behavior and session handling • Aligning chatbot responses with FAQ and support needs • Testing the assistant against common Tier-1 customer queries • Improving the web embed structure for cleaner deployment • Reviewing the implementation for basic frontend and security hygiene

Decision Logic The project was designed around a practical build-vs-buy decision. Option Pros Cons Decision Build custom NLP chatbot from scratch Full control, deeper ML ownership Slower, harder to maintain, unnecessary for narrow FAQ scope Not selected Use static FAQ page only Simple, low cost Poor user experience, no conversational flow Not enough Use Botpress webchat Fast deployment, maintainable, conversational interface Platform dependency Selected Botpress was the right choice because the business needed fast support automation, not research-grade NLP.