Skip to content

Update ChatGPT using OpenAI APIs.md #223

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Sep 25, 2024
Merged

Update ChatGPT using OpenAI APIs.md #223

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
32 changes: 5 additions & 27 deletions docs/4. Integrations/ChatGPT using OpenAI APIs.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,37 +4,22 @@

_**Leverage the capabilities of GPT models developed by OpenAI by using OpenAI APIs.**_

## Pre-requisites
1. Create an account in OpenAI platform [platform](https://openai.com/product)
2. Create new API keys
3. Make sure there is enough credit in OpenAI account to allow usage.
4. Pricing for API calls is as per OpenAI platform [pricing](https://openai.com/pricing)

<img width="1060" alt="Screenshot 2024-02-07 at 10 50 58 AM" src="https://github.com/glific/docs/assets/141305477/d3c22780-a4ae-4aaa-b196-1d079fce7eb5" />

_Above is a representative image from the OpenAI platform account_

## How it will work

1. Question is asked by the user after the flow is initiated
2. OpenAI API call is made using webhooks within the flow
3. Model being called to generate responses is `gpt-3.5-turbo-16k`
4. The response to the question is then sent to the user

## Steps to follow to set up OpenAI API in Glific platform
1. Go to `Settings`, find `OpenAI (ChatGPT) (Beta)`
2. Click on `is active` and paste the OpenAI API key from your OpenAI platform account

<img width="591" alt="Screenshot 2024-02-07 at 10 48 51 AM" src="https://github.com/glific/docs/assets/141305477/87adef8e-1858-4aa6-8b46-7f976cfbb9fb" />
3. Question, prompt to the model, and the type of model being called to answer the question or perform the task is provided inside the webhook params
4. The response is returned from the GPT models which can be sent to the user directly or used as desired in the flow.


## Using the webhook for OpenAI API call in a Glific flow

<img width="1069" alt="Screenshot 2024-02-07 at 9 59 24 AM" src="https://github.com/glific/docs/assets/141305477/babb9e27-fce5-4b2b-81d9-58017fd1a079" />
<img width="930" alt="Screenshot 2024-09-25 at 11 10 13 AM" src="https://github.com/user-attachments/assets/3307ead2-1dc1-4459-a31b-ec11579b475d"/>

_Representative image to explain the steps in OpenAI API calls in a simple flow_

[Sample flow](https://drive.google.com/file/d/1jl0NWn73YlK1qMRedaPnGCQwDKHfG_7j/view?usp=drive_link)

[Sample flow](https://drive.google.com/file/d/1r2dKE1i95lKrz6V1BiEyvZTXgxXgX87N/view?usp=sharing)

1. Get the user question
2. In `call a webhook` node, select `function` and paste function name as `parse_via_chat_gpt`
Expand All @@ -43,8 +28,6 @@ _Representative image to explain the steps in OpenAI API calls in a simple flow_

3. Share the following function body
` {
"contact": "@contact",
"results": "@results",
"question_text": "@results.question",
"gpt_model":"gpt-4o",
"prompt":"Provide an additional prompt for the model"
Expand All @@ -57,11 +40,6 @@ here `question_text` is the parameter name corresponding to user question.
<img width="671" alt="Screenshot 2024-05-22 at 3 52 18 PM" src="https://github.com/glific/docs/assets/141305477/09211b9a-132b-4964-9388-a1aa00ed28bc" />



4. The response from GPT is shown as `@results.webhookresultname.parsed_msg`, in the given example `gpt_response` is the webhook result name. (see the first image)


## Limitations
1. Text to speech and speech to text models cannot be used via this webhook call.

_Reach out to the Glific team to flag any further customizations within this functionality_
Loading