To learn stream processing, it helps to have a real-time data source to practice on. Unfortunately, there aren’t as many public sources for real-data data as there are for historical data. We plan to change that by hosting many real-time data sources ourselves which are easily accessible via our own quixstreams Python library or by public websockets.
| Name | Websocket Address | Summary |
|---|---|---|
| Customer support chat stream | wss://ai-chat-wss-demo-realtimedatasources-prod.deployments.quix.io/timeseries | A continuous stream of chat messages between virtual customers and virtual customer support agents who work at a fictitious electronics retailer. The conversations generated by LLama2 AI Agents. |
You can connect to this source with the Websocket client of your choice. Here are a few suggestions.
- Browser based: https://websocketking.com/
- Python: websocket-client
Each data source produces a distinct data structure and generates data in different ways. Here you can find more details on how each of the sources work:
This is a continuous stream of chat messages from interactions between virtual customers and virtual customer support agents who work at a fictitious electronics retailer. Both the customers and the support agents are powered by large language models (Llama2 via llama-cpp) hosted in serverless containers. The customers are prompted to ask for help about recent purchases using various tones of voice and the support agents are prompted to politely assist them.
This data source consists of 2 AI's using a Llama model from hugging face. One AI is a customer who is having a problem with a recently purchased electronic item. The other AI is a support agent at ACME Electronics.
The chats are available via this websocket API: wss://ai-chat-wss-demo-realtimedatasources-prod.deployments.quix.io/timeseries
Example response: Here is an example response:
{
"timestamp": 1702371373977514000,
"role": "agent",
"text": " Of course! I understand your concerns and frustrations regarding the defective washing machine you purchased from ACME Electronics. I apologize for any inconvenience this has caused, and I'm here to help resolve the issue as best as possible.\nThank you for informing me that you are not interested in pursuing a warranty claim or repair of the defective parts at this time. However, please note that we do have a return and refund policy in place for customers who are unhappy with their purchase. Under this policy, you may be eligible to receive a full refund for the purchase price of the washing machine, provided it is returned in its original condition within a certain timeframe.\nTo assist you further, could you please provide me with some additional information? Specifically, I would need to know your order number or proof of purchase so that we can locate your record and process your refund accordingly. Additionally, please confirm the date on which you purchased the washing machine and any other relevant details about the product.\nOnce I have this information, I will be happy to guide you through the return and refund process as quickly and efficiently as possible.",
"conversation_id": "ef9a56a2-9fb2-4bc9-a856-53a4d4ba951a",
"quix": "This data source is provided for free by Quix. Goto https://quix.io or the open source GutHub repo (https://github.com/quixio/real-time-data-sources) for more information.",
"datetime": 1702371373977
}- Explore the back end for this service in read-only mode: https://portal.platform.quix.io/pipeline?workspace=demo-realtimedatasources-prod
- See the source code for the services that generate the data: Agent and Customer