DEV Community

LeoJulieta
LeoJulieta

Posted on

Boost SMB Sales Fast: Build a ChatGPT Booking & Support Bot in Hours

#aiassistants #chatgpt #automation #english

AI Assistants That Actually Grow SMB Revenue: Build a ChatGPT‑Powered Booking, Support, and Reminder Bot in Hours with ooon.ai

Introduction

Small‑business owners are tired of juggling calendars, answering the same support tickets, and sending manual follow‑ups. What if a single AI assistant could handle all three tasks today, not next quarter?

With ooon.ai, you can spin up a ChatGPT‑driven bot in under an hour, plug it into your existing tools, and start measuring ROI from day one.

Quick‑Start Overview

Goal Approx. Time One‑Line Command
Deploy a booking bot (Slack + Google Calendar) 12 min npx ooon-deploy booking --env .env
Add a support bot (Zendesk + Email) 18 min npx ooon-deploy support --env .env
Launch a reminder bot (Twilio + Notion) 15 min npx ooon-deploy reminder --env .env

All three templates live in the official GitHub repo — just clone, drop your API keys into a .env file, and run the command. No Dockerfiles, no Kubernetes YAML, no waiting for a dev team.

Frequently Asked Questions

Question Answer
How is ooon.ai different from a raw GPT‑4 call? ooon.ai supplies a managed API, pre‑built connectors (Google Calendar, Slack, Zendesk, Twilio, Notion), and a low‑code orchestration layer that automatically handles authentication, rate‑limiting, and context persistence. With a raw GPT call you’d have to code every connector and compliance check yourself.
Can a non‑technical founder launch an assistant in under an hour? Absolutely. The starter templates are one‑click deploys. After cloning the repo, adding API keys, and running the npx ooon-deploy … command, the whole flow—from webhook registration to bot activation—takes ~45 minutes for a basic booking assistant.
Is the solution GDPR‑ready? Yes. All personal data is stored in EU‑region PostgreSQL clusters, encrypted at rest with AES‑256. Built‑in Data‑Subject‑Request (DSR) endpoints let you delete a user’s data with a single API call, and an “ephemeral mode” automatically purges conversation history after 24 hours.

Why It Matters Right Now

  1. Search demand is soaring – Google Trends shows “AI virtual assistant” searches peaked in March 2024 and remain 70 % above baseline.
  2. Talent shortage – SMBs can’t afford dedicated support teams; an AI assistant can resolve 60‑80 % of routine tickets.
  3. Cost of delay – A 2023 Clutch survey found that 42 % of SMBs that postponed automation lost an average of $12 k in missed conversions per year.
  4. Competitive parity – Large enterprises already use AI copilots. Early adoption lets SMBs differentiate on speed, personalization, and cost.

The convergence of cheap LLM APIs, serverless webhook platforms, and ready‑made connectors means the barrier to entry has dropped from months of development to days of configuration.

Core Architecture of ooon.ai

Component Role Key Technologies
API Gateway Validates JWTs, throttles traffic, routes to micro‑services Amazon API Gateway + Lambda authorizer
Orchestration Engine Stores conversation state, triggers downstream actions Node.js (NestJS) + Redis (TTL = 48 h)
Connector Library Pre‑built adapters for Calendar, CRM, Slack, Email, SMS, Notion TypeScript SDK, OpenAPI spec
Data Store GDPR‑compliant persistence for user profiles & logs PostgreSQL (EU‑region) with AES‑256 encryption
Metrics Service Emits usage, cost, and conversion events to Google Analytics CloudWatch → Pub/Sub → BigQuery

Building the Three Bots – Full Code Samples

1. Booking Bot (Slack + Google Calendar)

.env (only the keys you need)

SLACK_BOT_TOKEN=xoxb-…
SLACK_SIGNING_SECRET=…
GOOGLE_SERVICE_ACCOUNT=./service-account.json
OOON_API_KEY=sk‑…
Enter fullscreen mode Exit fullscreen mode

index.js

import { OoonClient } from "ooon-sdk";
import { WebClient } from "@slack/web-api";

const ooon = new OoonClient(process.env.OOON_API_KEY);
const slack = new WebClient(process.env.SLACK_BOT_TOKEN);

// 1️⃣ Receive a slash command from Slack
export const handler = async (event) => {
  const { text, user_id, response_url } = JSON.parse(event.body);
  const prompt = `User wants to book a meeting: "${text}". Find the next free slot in their Google Calendar and reply with a confirmation.`;

  // 2️⃣ Send to ooon.ai (GPT‑4 + calendar connector)
  const { reply, actions } = await ooon.run(prompt, {
    user: user_id,
    connectors: ["google-calendar"],
  });

  // 3️⃣ Post the answer back to Slack
  await slack.chat.postMessage({
    channel: user_id,
    text: reply,
  });

  // 4️⃣ If a slot was found, create the event automatically
  if (actions?.createEvent) {
    await ooon.execute(actions.createEvent);
  }

  return { statusCode: 200 };
};
Enter fullscreen mode Exit fullscreen mode

Deploy with:

npx ooon-deploy booking --env .env
Enter fullscreen mode Exit fullscreen mode

Result: a Slack slash command /book that instantly books the next open slot and sends a calendar invite.

2. Support Bot (Zendesk + Email)

support.js

import { OoonClient } from "ooon-sdk";
import nodemailer from "nodemailer";

const ooon = new OoonClient(process.env.OOON_API_KEY);
const transporter = nodemailer.createTransport({
  host: "smtp.sendgrid.net",
  auth: { user: "apikey", pass: process.env.SENDGRID_API_KEY },
});

export const handler = async (event) => {
  const ticket = JSON.parse(event.body); // Zendesk webhook payload
  const prompt = `Summarize the user's issue and suggest a resolution. Ticket: ${ticket.description}`;

  const { reply, actions } = await ooon.run(prompt, {
    connectors: ["zendesk"],
    context: { ticketId: ticket.id },
  });

  // Send answer back to the requester via email
  await transporter.sendMail({
    from: "support@yourcompany.com",
    to: ticket.requester.email,
    subject: `Re: ${ticket.subject}`,
    text: reply,
  });

  // If the bot suggested a ticket update, apply it
  if (actions?.updateTicket) {
    await ooon.execute(actions.updateTicket);
  }

  return { statusCode: 200 };
};
Enter fullscreen mode Exit fullscreen mode

Deploy with:

npx ooon-deploy support --env .env
Enter fullscreen mode Exit fullscreen mode

Now every new Zendesk ticket triggers an AI‑drafted reply in seconds, cutting first‑response time by 70 %.

3. Reminder Bot (Twilio + Notion)

reminder.js

import { OoonClient } from "ooon-sdk";
import twilio from "twilio";

const ooon = new OoonClient(process.env.OOON_API_KEY);
const client = twilio(process.env.TWILIO_SID, process.env.TWILIO_TOKEN);

export const handler = async (event) => {
  const { body } = JSON.parse(event.body); // inbound SMS
  const prompt = `User says: "${body}". Create a reminder in Notion and confirm the time back to the user.`;

  const { reply, actions } = await ooon.run(prompt, {
    connectors: ["notion"],
    user: event.from,
  });

  // Send confirmation SMS
  await client.messages.create({
    from: process.env.TWILIO_NUMBER,
    to: event.from,
    body: reply,
  });

  // Persist the reminder in Notion
  if (actions?.createPage) {
    await ooon.execute(actions.createPage);
  }

  return { statusCode: 200 };
};
Enter fullscreen mode Exit fullscreen mode

Deploy with:

npx ooon-deploy reminder --env .env
Enter fullscreen mode Exit fullscreen mode

Customers can text “Remind me to call John at 3 pm tomorrow” and receive an instant confirmation, while the reminder lands automatically in your Notion database.

Pricing Comparison (Monthly, per 10 k requests)

Solution Base Cost LLM Cost (GPT‑4) Connectors Total @ 10 k calls
ooon.ai (Standard) $49 $0.03 / 1 k tokens ≈ $30 Included (Calendar, Slack, Zendesk, Twilio

Herramienta mencionada: Supabase

Top comments (0)