Skip to content

WhatsApp and Telegram AI chatbot

Customer messaging for small service businesses · Interactive prototype with sample data, solo

Updated

Before

Staff answered every WhatsApp message by hand and copied hot leads into the CRM later.

After

The bot answers routine questions, scores each lead and hands hot or unsure chats to a person.

How it works
  1. 1Person:Customer

    Writes on WhatsApp or Telegram.

  2. 2Software:Bot

    Follows the published flow and answers from the knowledge base, in the customer's language.

  3. 3Software:AI

    Scores the lead as cold, warm or hot and writes a short summary.

  4. 4Person:Staff

    Takes over hot or doubtful chats from the shared inbox.

    Software:Service

    Sends the lead and summary to the CRM.

Always on

  • Below the confidence line, the bot hands the chat to a person.
  • Every AI reply shows the source it used.
  • The team sees every chat live in one inbox.

Problem

Customers wrote on WhatsApp and Telegram at all hours. Staff answered by hand, replies came late, and hot leads got lost in the chat.

Diagnosis

A bot that invents answers loses customers, so it replies only from the company's own documents and hands over when unsure.

Options I weighed

  1. How does the bot stay inside what the business knows?

    Chosen

    Answers from uploaded documents, with a handoff below a confidence line

    Every reply shows its source, and a doubtful one goes to a person. Cost: the team must keep the documents current.

    • Rejected

      A free-running GPT-4 prompt

      Sounds sure when it is wrong, and nobody can check where an answer came from.

  2. Who controls how a conversation goes?

    Chosen

    The team, in a visual flow builder

    Fixed steps for greeting, questions and handoff; AI fills only the open parts. Cost: flows need an owner.

    • Rejected

      Let the model lead every chat

      Conversations drift, and a lead can leave without being asked the key questions.

Solution

One system for both channels: a visual flow builder sets the conversation, each lead gets a score, and hot leads reach the CRM. Built on sample chats.

Interface
  • Next.js
  • React
  • Tailwind CSS
  • Radix UI
  • Recharts
  • GSAP
Delivery
  • Vitest
  • Cloudflare Pages

Related services