Adds AI features such as chat assistants, document summaries and smart search to your product, built on OpenAI models with human review.
Used for
- AI chatbots
- Document summaries
- Smart search
- Drafting replies
AI chatbot development
AI chatbots that answer from your content and know when to pass to a person
Support and sales chatbots for web, app and WhatsApp that answer from your own content and hand off to your team.
At a glance
The short version
- What you get
- Customer chat
- Agent console
- Knowledge and quality
- How it works
- Web widget, app or WhatsApp
- Chatbot backend
- Help articles and order APIs
- Chatbot
- 2 more
- Runs on
- Website
- Web app
- Android
- iOS
- Admin
- Built with
- and 2 more tools
- Who it's for
- Cost depends on
- Channels
- Live-data integrations
- Languages
How to start
Tell us what you need in your own words. You talk to the developers who would build it and get a written, line-item estimate.
Get a project estimateWhat separates a useful chatbot from a frustrating one
Most chatbot frustration comes from two failures: bots that invent answers, and bots that trap people in loops.
Every answer is retrieved from content you control or fetched live from your order, booking or CRM system. Payments, refunds and cancellations follow fixed flows the model cannot improvise. When confidence is low, the topic is sensitive or the customer asks for a human, the bot hands over with the conversation so far.
Challenges
Work a chatbot should take off your team
Repeat questions swamp the queue
Order status, returns and pincode questions whose answers already exist.
Replies slow down after hours and in sales
Response times stretch at night and in festive sales.
Customers write in Hinglish
Mixed Hindi and English in Roman script, with spellings keyword bots miss.
WhatsApp leads go cold
Enquiries wait unqualified until someone logs them in the CRM.
What is included
Inside the build
Platform: Web widget, app, WhatsApp
Customer chat
- Answers from help articles and policies, with source links
- Live order, delivery and refund status via your APIs
- Pincode serviceability and COD eligibility checks
- Lead capture with consent, synced to your CRM
Platform: Web app
Agent console
- Hand-off queue with transcript and customer context
- Suggested replies agents edit before sending
Platform: Admin dashboard
Knowledge and quality
- Sync articles, PDFs and FAQs; re-index on change
- Unanswered-question log that shows content gaps
- Evaluation set re-run on every prompt or model change
How it works
How a customer question becomes a sourced answer, or a hand-off
The bot answers what your content and systems can support, and passes everything else to your team with the conversation attached.
- Web widget, app or WhatsApp
Step 1: Customer asks in their own words
Questions arrive in English, Hindi or Hinglish, typos included. The bot works out what is being asked before it looks anything up.
- Chatbot backend
Step 2: Bot checks who is asking
For order or account questions it asks for an OTP or a login first, so personal details only reach the right person.
- Help articles and order APIs
Step 3: Answer is found in your content or systems
Policy questions are answered from your help pages; order, refund and delivery status come live from your own systems.
- Chatbot
Step 4: Reply is sent with its source
The customer gets a short answer and a link to the policy or order it came from. Refunds and cancellations follow fixed flows.
- Support agent
Step 5: Unsure or upset? A person takes over
Low confidence, a sensitive topic or a request for a human moves the chat to your agents with the full transcript.
- Admin dashboard
Step 6: Gaps are logged and fixed
Unanswered questions show which help articles are missing, and every fix is re-tested before the bot starts using it.
Then it starts again at step 1: Customer asks in their own words
Integrations
Channels and systems we connect
Channel
WhatsApp Business Platform
Service replies, approved templates and opt-in handling.
Orders
Shopify, WooCommerce or a custom OMS
Order lookup after OTP verification.
Logistics
Shiprocket, Delhivery and courier APIs
Live tracking status.
Helpdesk
Freshdesk, Zendesk, Zoho Desk
Hand-off inside your agents’ existing tool.
CRM
Zoho CRM, HubSpot, Salesforce
Qualified leads and chat summaries.
Payments
Razorpay payment links
Created by your backend, never by the model.
Model
OpenAI or open-source models
Chosen per language and cost.
Privacy & security
Keeping conversations safe
Verification before personal data
Order and account details appear only after OTP or login.
Masking in transcripts
Phone numbers, emails and addresses are masked in logs and, where possible, before model calls.
Prompt-injection defences
User text and retrieved content are treated as data; tools are limited to an allow-list.
Retention and DPDP rights
Configurable retention; DPDP Act 2023 erasure requests handled from the admin panel.
Platforms
Where the chatbot lives
A widget on your website, a chat screen inside your apps for Android and iOS, and a web console and admin panel for your team.
- Web
Website
Marketing and content websites: fast, search-friendly pages with a CMS your team can update without a developer.
- Web
Web application
Browser-based applications with logins, roles and workflows, such as customer portals, SaaS products and internal tools.
- Mobile
Android app
Apps for Android phones and tablets, tested on budget and mid-range devices and published on Google Play or privately.
- Mobile
iOS app
Apps for iPhone and iPad, built to Apple's guidelines and released through TestFlight and the App Store.
- Back office
Admin dashboard
Back-office panels for operations, support and finance teams: orders, users, content, reports and permissions.
Technology
Technology behind the chatbot, and why it matters
Language models understand questions and phrase replies; Python and Node.js services fetch facts from your systems; PostgreSQL and Redis store conversations and keep replies quick.
Connects your product to AI language models from several providers, so each task uses a suitable model and you can switch later.
Used for
- AI assistants
- Answers from documents
- Workflow automation
- Ticket triage
Backend
Python
A programming language for AI features, data processing and automation: the engine behind document reading, reports and smart search.
Used for
- AI features
- Document reading
- Reports and analytics
- Task automation
Backend
Node.js
Runs the server side of apps: fast, scalable back ends that power your app, website and integrations.
Used for
- App back ends
- Live order tracking
- Chat and notifications
- Payment processing
Data
PostgreSQL
A reliable database for the records your business runs on: orders, payments, bookings and stock, kept accurate and easy to report on.
Used for
- Orders and customers
- Payments and ledgers
- Stock and inventory
- Bookings
Data
Redis
Keeps frequently used data in fast memory, so apps stay quick on busy days and live features like order tracking feel instant.
Used for
- Faster apps
- Live order status
- Shopping carts
- Job queues and alerts
Web
Next.js
A modern web technology for fast, search-friendly websites, online stores and web applications that load quickly on mobile.
Used for
- Business websites
- Online stores
- Web apps
- Customer portals
Cloud & DevOps
AWS
Cloud hosting for your app, website and data, with data centres in India and room to grow when traffic rises.
Used for
- App hosting
- File and photo storage
- Backups
- Busy sale days
Plain-English glossary
Chatbot terms, in plain English
- LLM (large language model)
- The model that lets the bot understand a messy Hinglish question and phrase a clear reply. On its own it knows nothing about your return policy or a customer’s order, so we never let it answer those from memory.
- RAG (retrieval-augmented generation)
- Before replying, the bot searches your help articles and policies and passes only the matching paragraphs to the model. Update an article and the bot’s answer changes with it, with no retraining.
- Hallucination
- A confident answer that is wrong, like a delivery promise nobody made. Sourcing, checks and hand-off make it much rarer, but no chatbot is right every time, so sensitive topics always have a route to your team.
- Guardrails
- The limits we set around the bot: topics it declines, actions only fixed flows may perform, and checks that block replies without a source. They let the bot talk naturally without ever inventing a refund rule.
- Prompt injection
- A trick where someone types instructions such as “ignore your rules” to make a bot misbehave. We treat customer messages and retrieved text as data, never as commands, and limit which tools the bot can call.
- Customer service window
- On WhatsApp, the 24 hours after a customer’s last message, during which your bot may reply freely. Messages you start outside that window must use templates approved in advance.
Product preview
Screens your customers and support team use
Illustrative screens, using a support chatbot for an online store as the example.
Customer chat. Customers ask in their own words; order details appear only after OTP verification, and the bot hands over when a change needs a person. 
Agent console. Agents see why each chat was handed over, the customer’s language and the transcript, so nobody asks the customer to repeat themselves. 
Admin dashboard. Your team sees how many chats were answered from a source, how many went to agents and which questions found nothing, then decides what to write next. 
Admin dashboard. Articles, PDFs and FAQs are re-indexed when they change, and expired content can be switched off in one place so the bot stops quoting it.
Illustrative preview
Sample screens: names, prices and figures are examples, not client data.
Industries
Where support chatbots fit
Services
Services behind a chatbot build
AI app development
AI features and AI-first apps built on model APIs, retrieval over your own data, workflow automation or custom ML, with evaluation and human review.
Web application development
Browser-based products, customer portals, dashboards and internal tools, built on clean data models with secure roles and integrations.
API development
Secure, documented REST and GraphQL APIs, plus integrations that connect your apps to payment, logistics, GST, messaging and business systems.
Admin panel development
Back-office dashboards built around your team's daily tasks: order queues, approvals, catalogue management, payouts, reports and audit logs.
Mobile app development
Native and cross-platform Android and iOS apps, from first release through regular store updates, built by our in-house designers and engineers.
Work
Illustrative sample project
AI solutions
Illustrative sampleAI customer support assistant for a D2C brand
An illustrative AI assistant that answers order, return and product questions from a D2C brand's own policies and order data, with sources and a human hand-off.
- E-commerce
- AI assistant
Runs on
- Website
- Admin
Built with
- Next.js
- Python
- OpenAI APIs
- LLM integrations
- PostgreSQL
- +1 more
Cost drivers
What shapes a chatbot estimate
Channels
Web, app and WhatsApp each have their own formats, limits and tests.
Live-data integrations
Each system the bot reads adds API work, verification and tests.
Languages
Each language needs its own evaluation questions and a fluent reviewer.
Agent console scope
Your existing helpdesk is lighter than a custom console.
Conversation volume
Model usage is a running cost, reduced with caching and smaller models.
Estimates are written from your scope, with the effort and assumptions behind each line item.
How pricing worksFAQ
Frequently asked questions
Should our chatbot be rule-based or use an LLM?
Usually both. Fixed flows suit steps where wording and outcome must be exact: payments, cancellations, address changes, consent. A language model handles the open-ended part, such as understanding a Hinglish question, finding the right policy paragraph or summarising a complaint for an agent. We decide flow by flow, so the model never improvises a refund rule or a delivery promise.
Can the chatbot work on WhatsApp?
Yes, through the WhatsApp Business Platform, using Meta’s Cloud API directly or a business solution provider you already use. Within the customer service window the bot can reply freely; messages you initiate outside it need pre-approved templates and the user’s opt-in. WhatsApp supports short text, buttons and lists, so we design those flows separately from the web widget.
How long does a chatbot project typically take?
A support bot on one channel, answering from existing help content with one live-data integration such as order status, typically takes six to ten weeks including evaluation. WhatsApp, a custom agent console, extra languages or several backend systems extend that. Content is often the slowest part: contradictory or outdated policies need fixing before any bot can answer from them.
What happens when the bot does not know the answer?
It says so and offers a next step instead of guessing. If retrieval finds no relevant source, confidence is low or the customer asks for a person, the chat moves to an agent queue with the transcript attached. Outside support hours it raises a ticket instead. Each case lands in the unanswered log, which often points to a missing help article.
Will customer chats be used to train an AI model?
Not unless you decide so. With hosted model APIs we use settings and agreements under which your data is not used for provider training, and check each provider’s retention terms. Transcripts stay in your database with masking and a retention period you set. Fine-tuning on past chats later would be a separate, consented step with personal details removed first.
Next step
Share a week of real customer questions
Send an export of recent chats or tickets. We will show which a bot can answer from your content.