Skip to main content

The AI chatbot platform that answers from your data — on web, WhatsApp, and voice

Train it on your own documents. It answers on your website and WhatsApp, captures leads automatically, and escalates to a human agent when the question needs one.

Production-grade RAG · 4× retrieval accuracy · sub-300ms latency · 100% escalation delivery

PrismBot AI chatbot platform — admin analytics dashboard
retrieval accuracy vs naive search
<300msvector search latency
100%escalation delivery rate
0cross-tenant data leakage

Everything Your AI Chatbot Needs

Domain-Specific Understanding

Domain-Specific Understanding

Upload your PDFs, CSVs, TXT files, or a website URL. PrismBot indexes everything and answers from what you actually gave it — not from generic training data.

Conversational AI Chatbot

Conversational AI Chatbot

Multi-turn conversations with contextual memory and query expansion. A follow-up question about "that plan you mentioned" gets the right answer, not a confused restart.

Hosted Admin Interface

Hosted Admin Interface

Watch live conversations, manage your knowledge base, configure models, and take over any chat in one click — all from one dashboard.

Analytics & Insights

Analytics & Insights

See visitor counts, conversation totals, escalation rates, and captured leads. The analytics show what customers are asking and where the bot is falling short.

Web Widget

Add a conversational AI chatbot to your website

Three lines of embed code. That's the full integration. Copy them before your site's closing body tag and a fully branded chat widget appears — no backend changes, no developer sprint.

  • Copy-paste embed snippet — works on any website or CMS
  • Customise logo, brand colours, widget placement, and bot persona
  • Live in minutes once your knowledge base is uploaded
  • Multi-domain embed — one bot across multiple websites
3lines of code to go live
🌐

Website Embed

Any website · Any CMS · Any web app

<script src="/socket.io.min.js"></script>

<script src="/chatbot.js"></script>

<script src="/widget.js"></script>

Paste before your site's </body> tag — that's it.

11

colour theme controls

1

knowledge base, every channel

💚

WhatsApp Business API

Meta Graph API · Official integration

WhatsApp Phone Number ID

123456789012345

Business Account ID

987654321098765

Status

● Connected — webhook active

One-time setup. After that, WhatsApp automation runs fully automatically.

WhatsApp Channel

WhatsApp AI chatbot, powered by Meta's API

Your customers are already on WhatsApp. PrismBot connects to your WhatsApp Business number via Meta's official Graph API and turns it into a 24/7 meta AI chatbot on WhatsApp — same knowledge base as your website, no duplicate training.

  • Full WhatsApp Business API integration via Meta's official Graph API
  • Same knowledge base as your website — no duplicate training
  • CRM lead capture, escalation, and conversation history on WhatsApp
  • One-time Meta credentials setup. Most businesses are live within a few hours.
Set up WhatsApp AI chatbot →
Voice Channel

Voice and online chatbot AI on the same platform

Visitors speak, PrismBot responds in audio. The same online chatbot AI that powers your text chat handles voice from the same knowledge base — no retraining, no separate setup.

  • User speaks, bot responds in audio — ~3.3s to first audio via Sarvam AI
  • Barge-in interruption: visitor can cut the bot off mid-sentence
  • Outbound voice agent for appointment booking and lead qualification
  • Strong Indian accent support (en-IN) via Sarvam AI STT/TTS
~3.3sto first audio response
🎤

Voice AI Pipeline

Dedicated low-latency path

STT

Speech → Text

Sarvam AI · VAD included

RAG

Retrieval from KB

Direct RAG — no supervisor overhead

LLM

Answer generation

Streaming token delivery

TTS

Text → Audio

Sentence-by-sentence streaming

~3.3s

total end-to-end latency to first audio

Measurable Benefits

PrismBot AI chatbot platform admin dashboard showing live chat metrics

How PrismBot's AI chatbot platform works

From your first document upload to a live chatbot — the web channel needs no developer involvement.

01

Sign up & create your tenant

Create your account and set up a tenant. Your knowledge base, widget, and CRM data are isolated from everyone else from the start.

02

Upload your knowledge base

Drag-and-drop PDFs, TXT files, or CSVs through the dashboard — or paste a website URL to scrape. PrismBot takes care of chunking, embedding, and indexing.

03

Customise your bot persona

Set your brand logo, colour theme, widget placement, and a system prompt. The bot responds in your company's voice from the start.

04

Embed & go live

Copy three script lines onto your website. Connect your WhatsApp Business number if you need it. That's it — your AI chatbot platform is live.

PrismBot integrates with the tools you already use

WhatsApp Business API, OpenAI, Ollama for local LLMs, Sarvam AI for voice, and more — all configurable per tenant.

💬WhatsApp Business API
🤖OpenAI GPT-4.1
🦙Ollama (local LLMs)
🎙️Sarvam AI Voice
🗄️pgvector (PostgreSQL)
🍃MongoDB
🔌Socket.IO
📧AWS SES
🔍LangSmith Observability
🔐Google Auth / SSO
🌐Any Website (embed)
🔗Custom API
Admin Panel

Real-Time Chat Monitoring

PrismBot admin inbox — real-time AI chatbot conversation monitoring with one-click human takeover
Understanding AI Chatbots

What is an AI chatbot platform?

An AI chatbot is a software application that uses large language models (LLMs) and natural language processing to hold human-like conversations with users. Unlike rule-based chatbots that follow rigid decision trees, a conversational AI chatbot understands context, handles follow-up questions, and produces grounded, accurate responses from the knowledge you give it.

An AI chatbot platform is the complete infrastructure layer that lets a business deploy, manage, and improve AI chatbots across every channel. PrismBot is that platform. It ingests your documents, indexes them in a vector knowledge base, routes every user query through a production-grade RAG pipeline, and delivers answers on your website, WhatsApp, and via voice — with built-in lead capture, human escalation, and analytics — all from a single admin dashboard.

How does a RAG-powered AI chatbot stay accurate?

Most AI chatbots rely on a single vector search to find relevant content. PrismBot uses a three-stage hybrid pipeline: first, it expands every user query into three semantically diverse variants; second, it runs those variants through parallel vector searches and a BM25 keyword search simultaneously; third, it applies cross-encoder MMR reranking to select the five most relevant and diverse chunks before sending them to the LLM. The result is 4× better retrieval relevance compared to naive top-K retrieval. In practice: the chatbot gives the right answer, not a plausible-sounding one.

Online chatbot AI for customer support and lead capture

Businesses use an online chatbot AI to handle the repetitive inbound volume — FAQs, pricing questions, policy lookups — without scaling their support headcount. PrismBot pulls answers from your own knowledge base, escalates to a human agent when needed, and captures structured lead data automatically on every conversation.

What is the best AI chatbot for your business?

The best AI chatbot for any business comes down to four questions: Does it give accurate answers from your own data? Does it cover the channels you need — web, WhatsApp, voice? Can it run on your servers? And what does it actually cost? PrismBot is built to answer yes to all four — production RAG retrieval, three channels from a single deploy, full on-premises support, and free if you bring your own OpenAI key.

What does “AI chatbot online free” mean for PrismBot?

When PrismBot says free, it means the platform charges nothing when you supply your own LLM API key. You pay OpenAI directly — no platform fee, no markup. That makes PrismBot a practical option as an AI chatbot online for startups and SMBs that already have an API key. For teams who want CodePrism to manage LLM access, you pay the pass-through cost only.

For enterprise setups that need more than the standard platform — database connectors, ERP bridges, proprietary-system agents — explore our AI chatbot development services to see how CodePrism builds those integrations.

🧠

RAG-Powered Accuracy

Hybrid retrieval combines semantic search, BM25 keyword matching, and cross-encoder reranking to ground every answer in your actual documents. That's 4× better relevance than single-vector search.

🔒

Enterprise-Grade Security

Each tenant gets isolated storage, AES-encrypted API keys, and content guardrails as standard. On-premises deployment means your data stays inside your own infrastructure.

📊

Built-in CRM Agent

Every conversation captures lead signals automatically — name, email, intent — then asks one targeted follow-up. No manual data entry.

🌐

Omnichannel from Day One

One knowledge base covers your website, WhatsApp, and voice channel. No duplicate training, one admin panel.

Enterprise AI chatbot security

Tenant isolation, AES-encrypted API keys, content guardrails, and on-premises deployment. These are defaults, not add-ons.

🏢

Full Tenant Isolation

Each organisation gets its own knowledge base, vector namespace, chat history, and CRM data. Cross-tenant isolation is enforced at the database layer — it's a technical constraint, not a policy.

🔐

AES-Encrypted API Keys

Your custom OpenAI API key is encrypted with AES at rest inside PrismBot. LLM costs are billed directly to your OpenAI account — CodePrism never sees your spend.

🖥️

On-Premises Deployment

Run PrismBot entirely within your own infrastructure. Your knowledge base, embeddings, and conversation data stay in your environment — useful for regulated industries.

🛡️

Content Guardrails

PrismBot blocks jailbreak attempts, prompt injection, model role-switching, and oversized message attacks before they reach the LLM pipeline.

🚫

Zero LLM Training on Your Data

PrismBot never uses your documents or conversations to train any LLM provider's base model. Your knowledge base stays within your account boundary.

📋

Full Traceability

Every conversation, escalation, and CRM update is logged with timestamps and agent attribution. LangSmith tracing shows exactly what happened at every LLM call.

Choose any model. No lock-in.

PrismBot's embedding and generation layers are independently configurable. Switch from OpenAI to a local Ollama model in a single dashboard setting.

OpenAI GPT-4.1Generation & reasoning
OpenAI GPT-4.1-miniQuery expansion (cost-efficient)
Ollama (local)On-prem LLM — no cloud required
nomic-embed-textLocal embeddings via Ollama
Sarvam AI STT/TTSVoice — en-IN + Indian languages
OpenAI text-embeddingCloud embeddings (configurable)

Switching models is a configuration change, not a rewrite. PrismBot's embedding layer (OpenAI or Ollama) and generation layer (any OpenAI-compatible endpoint) are independently configurable per tenant. No vendor lock-in on LLM costs.

How PrismBot compares to other AI chatbots

Most AI chatbot tools use basic vector search and scripted escalations. Here's what changes when you use PrismBot instead.

FeatureChatbot.comWonderchatTidioPrismBot
Retrieval methodRule-based + basic AISingle-vector searchScript flows + AIHybrid RAG (semantic + BM25 + reranking)
Answer accuracy vs naive retrievalBaseline4× improvement
Built-in CRM lead captureExternal tool✓ Auto on every conversation
WhatsApp Business APILimitedLimited✓ Native (Meta Graph API)
Voice channel✓ STT + TTS (~3.3s latency)
On-premises deployment✓ Full self-hosted option
Model agnostic (no lock-in)OpenAI onlyOpenAI onlyOpenAI only✓ OpenAI + Ollama (configurable)
Real-time human takeoverBasic redirectBasicLive chat✓ One-click + push notifications

Use cases for an AI chatbot platform

Customer support, lead capture, internal knowledge bots, outbound voice agents — PrismBot handles each from the same platform.

🎧

Customer Support AI Chatbot

Deploy on your website to handle product FAQs, troubleshooting, returns, and order status from your own documentation — with instant human escalation for complex cases.

Ideal for: E-commerce, SaaS, Financial Services, Telecom

💚

WhatsApp Business Assistant

Turn your WhatsApp Business number into a 24/7 AI agent. It handles queries, qualifies leads, and books appointments — on the channel your customers already use.

Ideal for: Real estate, Insurance, Healthcare, Retail, Education

📚

Internal Knowledge Bot

Employees query HR policies, IT runbooks, compliance SOPs, and documentation in natural language — instead of searching SharePoint or asking colleagues.

Ideal for: Enterprise IT, HR, Legal, Compliance teams

🎯

AI Sales Lead Capture

Handles inbound queries, qualifies leads with targeted follow-up questions, captures contact details, and hands off to sales — 24/7, on web and WhatsApp.

Ideal for: SaaS, Financial Services, Real Estate, EdTech

🎤

Voice AI Assistant

Visitors speak naturally; the AI responds in audio within ~3.3 seconds. No typing needed — works on any website.

Ideal for: Healthcare, Automotive, Hospitality, Accessibility

📞

Outbound Voice Calling Agent

AI makes outbound calls for appointment confirmation, lead qualification, and payment reminders. It saves CRM data automatically and transfers to a human when asked.

Ideal for: Healthcare, Financial Services, EdTech

Pricing

Pricing — start free, online

Bring your own OpenAI API key and the AI chatbot platform costs nothing. You pay OpenAI directly — CodePrism charges zero platform fee and earns nothing on your LLM spend.

For teams who want CodePrism to manage LLM access, you pay the pass-through cost only. Formal subscription tiers are in development. Get in touch to discuss your deployment.

Self-hosted deployment available · White-label option available

FAQ

Frequently asked questions about PrismBot

Common questions about setup, security, pricing, and capabilities. Still stuck? Our team replies in under 2 hours.

PrismBot is a multi-tenant AI chatbot platform that trains on your own documents and answers on your website, WhatsApp, and via voice from a single admin dashboard. Most chatbots use scripted flows or a basic single-vector search. PrismBot runs a 3-stage hybrid RAG pipeline instead: query expansion, parallel semantic search combined with BM25 keyword matching, and cross-encoder reranking. That gets you 4× better answer relevance. It also qualifies leads automatically on every conversation — something most chatbot tools skip entirely.
PrismBot works on both — your website and WhatsApp — from a single platform. The same knowledge base, CRM agent, and admin dashboard serve your website chat widget and your WhatsApp Business number simultaneously. Add PrismBot to any website with a copy-paste embed snippet; on WhatsApp, connect your Meta Business API credentials through the dashboard and PrismBot becomes a full WhatsApp AI chatbot powered by Meta's official Graph API.
For enterprise deployments, three things actually matter: retrieval quality, data isolation, and deployment options. PrismBot uses hybrid RAG retrieval rather than basic vector search, enforces strict tenant isolation at the database and vector-search layer with zero cross-tenant leakage, and supports full on-premises deployment so nothing leaves your infrastructure. Add real-time escalation with 100% push-plus-email delivery, LangSmith LLM observability, and per-tenant AES-encrypted API keys — none of which you'll find in a generic chatbot builder.
Every tenant's knowledge base, embeddings, chat history, and lead data is fully isolated — enforced at the vector-database and query layer via namespace isolation in pgvector, not just by policy. Your custom OpenAI API key is encrypted with AES at rest. PrismBot supports on-premises deployment, so your data never has to touch a third-party server. Active content guardrails block jailbreak attempts, prompt injection, model role-switching, and oversized messages before they reach the LLM.
Yes. Bring your own OpenAI API key and the PrismBot platform costs nothing to use as an AI chatbot online. You pay only the actual LLM cost, billed directly to your OpenAI account. CodePrism adds zero markup. For teams who want CodePrism to manage LLM access, you pay the pass-through cost only. Formal subscription tiers are in development.
Accuracy is a pipeline design problem, not just a prompt problem. PrismBot generates three semantically diverse query variants from every user question, runs parallel vector searches on all three, merges results with BM25 keyword search, then applies cross-encoder MMR reranking to select the five most relevant, contextually coherent chunks from your knowledge base. This 3-stage approach delivers 4× better retrieval relevance than naive top-K retrieval, and every answer cites its source documents so you can verify accuracy. Multi-turn contextual memory ensures follow-up questions are handled coherently.
If your content is ready, a web chatbot can be live in minutes: upload documents through the dashboard, configure the widget, paste three script tags into your HTML. No developer sprint, no backend changes. WhatsApp takes a bit longer because of the one-time Meta credentials step, but most teams finish the full setup within a few hours once those credentials are in hand.
Yes. PrismBot integrates with WhatsApp via Meta's official Business API (Graph API). You supply your Meta access token, WhatsApp Phone Number ID, and Business Account ID through the PrismBot dashboard, and the bot immediately begins handling inbound WhatsApp messages from your knowledge base — qualifying leads, escalating to human agents when needed, and maintaining full conversation history in the same admin inbox you use for website chat.

Still have questions?

Talk to our team →
Product Demo

PrismBot in Action

Demo Video
--:--
0:00 / 0:00
PrismBot
PrismBot AI chatbot demo video preview

Ready to deploy PrismBot in your business?

Tell us your use case. We won't walk you through slide decks — we'll demo PrismBot on your actual knowledge base. Most businesses are live within a day.

Need a custom integration beyond the standard platform? Explore our AI chatbot development services →

PrismBot character