Use of AI sources

Are there any ways to use or implement open source or close sourced ai tools?

Hi - yes any AI tool that has a REST based interface can be used with Jigx and there are many examples with the OpenAI ChatGPT tools. Please let us have a little more context on what you would like to achieve and we can answer this in more detail or provide samples

1 Like

Okay so Jigx is compatible with REST based interfaces, thank you. Essentially I would like the ability to use an ai interface to scan pictures and videos and create captions. I know there is a way under media form you could download or add the picture onto whatever app you create, but if there is a way to use ai to then scan whatever is downloaded and it comes up with a caption that is what I am trying to do?

Hi - yes - this is possible and we have a demo scenario which in our scenario has a store employee take a photo of the store from the outside. The AI then analyses the image to make sure that store conforms to certain standards. We prompt the AI with prompts similar to the below. Let us know if you have something similar in mind and we can give you sample YAML for this scenario

"model": "gpt-4o",
  "response_format": {
    "type": "json_object"
  },
"messages": [
  {
    "role": "system",
    "content": "We are comparing a photo of the outside of a store to determine if a list of requirements have been met."
  },
  {
    "role": "system",
    "content": "The requirements are: all signs in blue and in good quality. \"The store logo\" should be visible in the window. Opening time 8am to 6pm should be visible in window. No trash cans should be visible in front of the store. Employees should stand outside wearing blue store shirts and black pants. \"Open XX hours\" should be visible. \"100% pure cotton\" should be visible.  look at the attached photo and return a list of all requirements and if they are being met or not, why or why not and if not what can be done to correct them."
  },
  {
    "role": "system",
    "content": "Analyze the photo and return a json array indicating which of the requirements have been met."
  },      
  {
    "role": "system",
    "content": "the response must be a json array of the following objects. {\"result\":[{\"requirement\": \"Description of requirement\", \"status\": \"status if the requirement has been met\", \"reason\": \"reason why the requirement has been met or not met\", \"correctiveSuggestion\": \"suggestion of how to meet the requirement.\"}]}"
  },   
  {
    "role": "user",
    "content": [
      {
        "type": "image_url",
        "image_url": {
          "url": image
        }
      }
    ]
  }

Yes I would like to do something very similar. I would like the preprompt on the back end to make sure the picture or video conforms to social media rules and standards. If that is possible in YAML that would be great!

All good. We will put a simple sample together and send to you. You will need to change the key in the function call to your own ChatGPT API key and that should give you what you need. More to follow

Hi - just letting you know a sample is on the way and we have had many people asking about this. We will include an example of image recognition as well as chat functionality with AI. Examples will include the ability to prompt your favourite AI backend. Very impressive results being seen