Configure ChatGPT in Zoho Cliq | Now with GPT-4!

Configure ChatGPT in Zoho Cliq | Now with GPT-4!

Updated for GPT-4 Support: The post and scripts has been revised to take advantage of the new GPT-4 model's capabilities. Please use the updated version for enhanced performance and accuracy.  

If you have been on the internet for the past few months, you have probably heard of ChatGPT. It has been making waves across the world as the next big technological revolution in artificial intelligence. ChatGPT is an advanced language model developed by OpenAI. It uses deep learning techniques and is trained on a massive amount of data to generate human-like responses to text inputs. With its ability to understand context, recognize patterns and relationships, and generate natural language, ChatGPT is capable of providing informative and engaging answers to a wide range of questions on diverse topics.

According to a recent PWC report, 67% of business executives think integrating AI into their business will greatly improve performance and efficiency. You can also integrate ChatGPT into Cliq by following these easy steps:
  • Create a bot with channel participation permission (make sure to check the options for the bot to "send messages" and "listen to messages" in the sub actions)
  • Go to Profile → Bots & tools → Bots → Create Bot → Enter the name, description and enable channel participation → Save.
A helpful rule of thumb is that one token generally corresponds to ~4 characters of text for common English text. This translates to roughly ¾ of a word (so 100 tokens ~= 75 words).

Message Handler:

If you want to ask questions to the bot in a one-on-one chat directly, you will have to modify the message handler. To do that, follow these steps:
  • Navigate to the message handler and click on Edit Code. Now copy the below code and paste it:
  1. info message;
  2. response = Map();
  3. // Need to add openAI token
  4. token = "Bearer sk-GlHHcXXXXXXXXXXXXXXXXXXXXXXXXX";
  5. header = Map();
  6. header.put("Authorization",token);
  7. header.put("Content-Type","application/json");
  8. contentList = list();
  9. contentList.add({"role":"user","content":message});
  10. params = {"model":"gpt-4","messages":contentList,"temperature":0.9,"max_tokens":2048,"top_p":1,"frequency_penalty":0,"presence_penalty":0,"stop":{"Human:","AI:"}};
  11. // Making post request 
  12. fetchCompletions = invokeurl
  13. [
  14. url :"https://api.openai.com/v1/chat/completions"
  15. type :POST
  16. parameters:params + ""
  17. headers:header
  18. detailed:true
  19. ];
  20. info fetchCompletions;
  21. if(fetchCompletions.get("responseCode") == 200)
  22. {
  23. // Populating the response to human readable format
  24. answer = fetchCompletions.get("responseText").get("choices").toMap().get("message").get("content");
  25. info "answer" + answer;
  26. response.put("text",answer);
  27. }
  28. else if(fetchCompletions.get("responseCode") == 401)
  29. {
  30. response = {"text":fetchCompletions.get("responseText").get("error").get("message")};
  31. }
  32. else if(fetchCompletions.get("responseCode") == 429)
  33. {
  34. response = {"text":fetchCompletions.get("responseText").get("error").get("message")};
  35. }
  36. else if(fetchCompletions.get("responseCode") == 503)
  37. {
  38. response = {"text":"Service Temporarily Unavailable"};
  39. }
  40. else
  41. {
  42. response = {"text":"I dont have any knowledge in this. Please ask me something else"};
  43. }
  44. return response;
  • Then navigate to this link
     and generate a token in openAI and then replace the token in line 4 (It should look something like this : "Bearer sk-K4ilep5NLxxxxxxxxxxxxxxxxxxxxxxxxxxx").
    Note: Project-based API keys from OpenAI will not work for this integration. You need to use a secret API key associated with your personal or organizational OpenAI account. 
  • Save the message handler. Now the bot is ready to answer your questions.

Participation Handler:

If you want to add the bot to a channel so it can answer questions from any participants, you will have to modify the participation handler. To do that, follow these steps:
  • Add the bot to the required channel.
  • Navigate to the bot participation handler and click on Edit Code. Now copy the below code and paste it.
  1. response = Map();
  2. if(operation == "message_sent")
  3. {
  4. info data;
  5. if(data.get("message").get("type") == "text")
  6. {
  7. response = Map();
  8. // Need to add openAI token
  9. token = "Bearer sk-GlHHXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX";
  10. header = Map();
  11. header.put("Authorization",token);
  12. header.put("Content-Type","application/json");
  13. contentList = list();
  14. contentList.add({"role":"user","content":data.get("message").get("text")});
  15. params = {"model":"gpt-4","messages":contentList,"temperature":0.9,"max_tokens":2048,"top_p":1,"frequency_penalty":0,"presence_penalty":0,"stop":{"Human:","AI:"}};
  16. // Making post request 
  17. fetchCompletions = invokeurl
  18. [
  19. url :"https://api.openai.com/v1/chat/completions"
  20. type :POST
  21. parameters:params + ""
  22. headers:header
  23. detailed:true
  24. ];
  25. info fetchCompletions;
  26. if(fetchCompletions.get("responseCode") == 200)
  27. {
  28. // Populating the response to human readable format
  29. answer = fetchCompletions.get("responseText").get("choices").toMap().get("message").get("content");
  30. response.put("text",answer);
  31. }
  32. else if(fetchCompletions.get("responseCode") == 401)
  33. {
  34. response = {"text":fetchCompletions.get("responseText").get("error").get("message")};
  35. }
  36. else if(fetchCompletions.get("responseCode") == 429)
  37. {
  38. response = {"text":fetchCompletions.get("responseText").get("error").get("message")};
  39. }
  40. else if(fetchCompletions.get("responseCode") == 503)
  41. {
  42. response = {"text":"Service Temporarily Unavailable"};
  43. }
  44. else
  45. {
  46. response = {"text":"I dont have any knowledge in this. Please ask me something else"};
  47. }
  48. return response;
  49. }
  50. }
  51. return Map();
  • Make sure to replace the token in line 10 with your Open AI token. 
  • Save the bot participation handler. That's it. You can now ask questions directly to the ChatGPT Bot in the configured channel without leaving Cliq.
Note: It should be noted that, if you want to use the bot at the organization/team level, it's better to use connections so each user can use their own openAI account instead of all the queries going through a single openAI account token. This approach can be beneficial as it can prevent one user's actions from negatively impacting the entire team's access to the API.