We’ve launched a new low-code screen template that utilizes the Fliplet AI API to empower you to use AI to summarize content. This sample screen will help users summarize a news article. You can modify the data or adjust the AI prompt in the developer options JS code.
How do I use this screen template?
Use this with any data set that you would want to let users summarize - news, reference material, or more.
What does this screen template do?
This screen uses the List from Data Source to show news. However, in the detail view there is a ‘Generate Summary’ button that when clicked will send the news content to AI to generate a summary that is then shown to the user.
How do I customize this screen template?
You can use this with your data, or use the developer options (the </> icon on the far right menu in Edit mode) and modify the JS code, including the AI prompt.
Interesting. So for example, if a user wants to have a quick summary on say ‘FSQS’ supplier verification, as part of a management systems action, then this screen could provide that?
Yes Kenneth! By default it will give you a two-paragraph summary of all the data provided. So you may want to adjust the prompt to make it more specific or relevant, but I’d encourage you to test it out first and see.
So. Does the summarised data need to be in fliplet tables first, or can I specify a url to go to? So for example, link to Nexis Lexis as a data source? Or API link to another cloud platform full of content? Very interesting for possible gap analysis.
The data does need to be in a Fliplet data source. Since we are making the calls to OpenAI in our Javascript we have to directly access the data and OpenAI doesn’t have the ability to access the web. You’d have to give it raw data in some format, but instead of putting it into Fliplet directly, you could look into syncing the data from Nexis Lexis into a Fliplet data source with an API or a Zapier integration.
If you modify the data in the data source you can try the summary as-is. If you want it to summarize something specific or in a different way, you may need to modify the AI prompt and potentially the data we send to AI in the screen JS.
Potentially, a community of people who specialise in management systems, when they find useful/interesting materials, that is ripped down to text and saved in a bucket somewhere. Then we search on that bucket? Using sources like Lexis is a problem as its highly structured, to make it easier for MK1 neural networks to interrogate. What fun!
We have now an interogation screen ‘Bot Marcus’ that goes to a seperate Chat GPT account for Business Compass, not the Fliplet Key.
Tested with management systems terminology and seems to handle it… Now exploring how this can enhance overall app usability for people new to the subject.
Hey Kenneth, great to here you got a version of it working! We have been considering a chatbot/doc analysis tool and were thinking we would possibly have common prompts as an option for users to select to start a conversation and/or refine the prompt more in the code so that it is always a bit more tailored to a specific subject. Do you think either of those would fit what you are looking to do?
Hi Brett, let us show what what has been achieved with dialogue so far and the plan. Sounds like there is common interest. I need my colleague who is also in France. Suggest some days/times to fo a call.
