Since ChatGPT was first released to the public in 2022, generative AI has become a public obsession. Within two months, ChatGPT became the fastest growing app in history, with 100 million users by January 2023.
On the back of this success, different companies have turned their hand to creating generative AI solutions, including Google’s Gemini, Midjourney and Stable Diffusion.
What is generative AI?
Generative AI refers to AI technology that uses deep learning (DL) to create text, images, and code based on the data or knowledge base they are trained on. Modern generative AI, such as ChatGPT, can generate high quality content from scratch.
What is deep learning?
Deep learning is a branch of AI that teaches computers to process data by linking neural networks, much like a human brain. In this way, AI can learn from being exposed to large data sets to process complex patterns and categorise unlabelled data.
The difference between generative AI and conversational AI
Conversational AI uses natural language processing (NLP) and machine learning (ML) to understand human language and respond in a natural, human-like way. Most virtual assistants and chatbots use this technology to better mimic human conversation, with back-and-forth messages that feel more organic to the user.
Where conversational AI is focussed more on human-like interaction with the user, generative AI is more suitable for generating unique content (images, text and code) based on the data it has been trained on.
What is a ChatGPT chatbot?
A ChatGPT chatbot is a bot that leverages the power of OpenAI and ChatGPT. Using this generative AI technology and linking to a knowledge base enables anyone to create a fully functioning AI chatbot without the need to know code or bot frameworks.
Why AI-powered chatbots are superior to rules-based bots
AI backed chatbots, like a ChatGPT chatbot, have technology and knowledge bases behind them that allow them to think like a human chat agent. Here are a few ways in which AI chatbots work better than rules-based bots:
| Rules Based Chatbot | AI Powered Bot |
|---|---|
| Follows a preset dialog tree | Can reply in a human-like way and understand context |
| Must be manually updated when information changes | Learns the more it interacts with users |
| Can only answer frequently asked questions it has been programmed to look for | Can respond to any query based on the knowledge base it is trained on |
| Must be programmed to understand and respond in different languages | Knows multiple languages out of the box |
Create your own chatbot with ChatGPT and ThinkAutomation
In six simple steps and minimal coding knowledge you can create your own ChatGPT chatbot using ThinkAutomation. Follow the steps below to build your own:
Step 1: Download and install ThinkAutomation
Download ThinkAutomation to any Windows 11 or Windows Server computer (on-premises or in the cloud).
Step 2: Create an OpenAI account
After you’ve signed up for an OpenAI account, obtain your OpenAI API key by logging in to the developer platform. Find and click on ‘API keys’, then click ‘Create new secret key’.
Step 3: Add your OpenAI API key
Copy and paste the OpenAI API key you’ve just created into ThinkAutomation. You can add it in the File – Server Settings – Integrations tab.
Step 4: Open the Knowledge Store browser
Add your business specific knowledge articles to the knowledge store (by importing PDF, Word, Text, HTML or Markdown documents). You can also set up a separate automation to populate your knowledge store via incoming emails, databases, or local files.
Step 5: Add a new Web Chat Message Source in ThinkAutomation
Click the New Message Source button and select Web Chat as the Message Source type. Follow the wizard and make sure to enable the ‘Use ChatGPT To Respond To Incoming Messages’ option.
Step 6: Your bot is ready!
Add the bot URL to your website or use the local URL internally. You can add more articles to your local knowledge store to improve the chatbots responses.
How much does OpenAI cost?
Creating an OpenAI account is free, but it uses a token-based pricing system where you’re charged for what requests your API uses. For example, API calls using GPT-4o are charged $0.0025 for every 1,000 input tokens and $0.01 for every 1,000 output tokens.
Benefits of building a chatbot with ChatGPT and ThinkAutomation
Building a ChatGPT chatbot using ThinkAutomation gives you several advantages over creating one just using ChatGPT. You’ll benefit from:
- Not having to train your own ChatGPT model. This is much more cost effective, and you don’t have to keep re-training your bot whenever your local knowledge store changes!
- On-premises knowledge. Your knowledge store is maintained locally on-site, so you don’t need to pay a 3rd party service (you’ll just need an OpenAI account).
- Not uploading potentially confidential or sensitive information. Your ChatGPT chatbot can still be hosted publicly via the ThinkAutomation Gateway, but this is just the chat form HTML. The ChatGPT integration and knowledge search is handled by your local ThinkAutomation server, so knowledge base articles are not stored by ChatGPT itself.
- Using the same knowledge store to create multiple bots. Reply to incoming chats, emails, SMS messages or API calls with a standardised bot response.
- Simple implementation. No technical knowledge of bot frameworks is needed.
- Ability to instantly improve bot responses. Simply add more articles to your knowledge store.
- Local storage of bot question and answer sessions. ThinkAutomation stores messages in a local database for future reporting.
- Added context from your own database. In addition to your knowledge store, you can read information from your own database. This allows real-time bot responses, such as ‘Do you have {product} in stock?’.