Sections
GenAI in SnapApp
on 07-17-2024 12:00 AM by SnapApp
73
Generative AI (GenAI) is a subset of artificial intelligence focused on creating new content, such as text, images, music, or code, by learning patterns from existing data.
SnapApp uses GenAI to generate content powered by Vertex AI using the data stored in SnapApp. It can generate comments, replies. It can also summarize documents and search them using human-like queries. It can also be used to generate code snippets and images.
Table of Contents
- GenAI in SnapApp
- Generate Text in UI
- Generate Image in UI
- Generate Text in expression
- Generate Text with Image in expression
GenAI in SnapApp
.png)
- Text and Image Generation using Generative AI
| Feature | Description | Access Point | Availability | Activation Point | Model Used |
|---|---|---|---|---|---|
| Generate Text (UI) | Text generation using Generative AI (genAI) in SnapApp involves creating coherent and contextually relevant text based on a given input or prompt. | Generate Text link which pops up the prompt space to apply the prompt message. |
Feeds, Pages, Knowledge Base Builder | Text/ Long Text Field settings | Text-bison@001 model of Vertex AI (supports 8000 tokens) |
| Generate Image (UI) | Image generation using Generative AI (genAI) in SnapApp involves creating new images from scratch or transforming existing ones by learning patterns from large datasets. | Page Builder | Image Field by changing image input mode. | OpenAI Image Generation model | |
| Generate Text (Expressions) | Text generation using Generative AI (genAI) in SnapApp involves creating coherent and contextually relevant text based on a given input or prompt. Here custom expression functions are used. | Everywhere as expression | From expression builder | Text-bison@001 model of Vertex AI | |
| Generate Text with Image (Expressions) | Text generation with images using Generative AI (GenAI) in SnapApp refers to the process of creating coherent and contextually relevant text descriptions or narratives based on the content of an image. Here custom expression functions are used. | Everywhere as expression | From expression builder | Gemini 1.5-pro model of Vertex AI | |
| Generate Image (Expressions) | Image generation using Generative AI (genAI) in SnapApp involves creating new images from scratch or transforming existing ones by learning patterns from large datasets. Here custom expression functions are used. | Everywhere as expression | From expression builder | OpenAI Image Generation model |
#### Generate Text in UI
Generate text by providing prompt message in case of Text and Long Text fields.
.png)
#### Generate Image in UI
Generate image by providing prompt message in case of changing input in image field.
.png)
#### Generate Text in expression
Generate text by providing expression function (GENERATE_TEXT()) from expression builder.
.png)
#### Generate Text with Image in expression
Generate text by providing expression function (GENERATE_TEXT_WITH_IMAGE()) from expression builder.
.png)
- Feeds using Generative AI (in peek view)
| Feature | Description | Model Used |
|---|---|---|
| Generative Summary | Context-based generative summary of the conversation. | Text-unicorn model (supports 32000 tokens) |
| Feed Autocomplete | Context-based auto-complete of reply, directly into the text-box | Text-unicorn model |
| Quick reply options | Context-based user aware quick reply options on actual valuable context. | Text-unicorn models |
.png)
In the above example, a summary of the conversation and relevant automatic feed is generated using GenAI.
- Generative Search (Native)
Generative search (native) is a GenAI driven search tool, which works by creating SQL select queries by understanding the question and also provide insights about the retrieved data.
It is SnapApp’s own generative search using GenAI features in the framework of SnapApp’s own IP and integrated with SnapApp. It uses Gemini-1.5-flash to query that supports one million tokens, gather insights and form answer to the questions. It is suitable for data related queries, filtering, sorting, doing aggregation and schema-based queries but does not support keyword-related search. It does not pass all the data into the model rather than only the resultant data set received from SQL Query.
.png)
In the above example, a summary of the anime with maximum episodes is searched, and on the basis of table records of SnapApp, it searched the appropriate record and produce the summary.
- Generative Search (Agent Builder)
Generative search Plus is a tool that uses google’s search and conversation API to create a datastore and answers question from the datastore.
It is a vector search which supports keyword related searches but unable to filter, sort and provide answers to schema-based questions.
To use this, you have to upload the table data to Google’s data store, then after populating those data, you can search. It basically search for the top 5 most relevant similar meaning word and in addition it produces the generative summary of the answer(preview feature).
.png)
The above example shows the top 5 related search with a general summary of the question, and each time you refresh, the summary will be changed.
- Document Search (Agent Builder)
DocSearch leverages Google’s search and conversation capabilities to sift through document content and stored within SnapApp.
It produces a summarized answer to the question along with proper documents and relevant snippets for supportive data reference. Tapping on the summary will open the relevant page of the document. This result also includes the record of the SnapApp and the supported documents.
.png)
In the above example, a relevant summary is generated on the basis of the question with relevant refential citations for the data.
Thank you for following these steps to configure your SnapApp components effectively If you have any questions or need further assistance, please don’t hesitate to reach out to our support team. We’re here to help you make the most out of your SnapApp experience.
For support, email us at snapapp@bluevector.ai
