# Introduction Welcome to UnifiedStacks documentation. # Platform Overview **Folders Tab** Let’s start by creating a new folder in UnifiedStacks dashboard. Click on the “New Folder” button in the top right corner of the dashboard. A folder is a container where you can manage multiple applications. You can view all the folders created by you as well as shared with you.
### Apps tab You can manage all your applications from the apps tab. Click on the “New App” button in the top right corner of the dashboard to create a new app. You can view all the applications created by you as well as shared with you.
### Templates You can create a blank app or choose from a list of pre-designed applications depending on your use case.
### Settings You can manage your plans and check usage from settings page.
### Application Builder Once you create a new application, you will see an interface with four sections.
1. Top bar - You can edit the name, preview, save and publish the application from the top navigation bar. 2. Left side - App Navigation bar. 3. Workflow - Drag and drop workflow elements in the section. 4. Workflow elements - List of available elements for app builder. ### Knowledge Base You can upload, review and manage documents and information relevant to the application from Knowledge Base page.
### Exports Once you publish an application, you will get an URL link which you can share with internal as well as external users.
### Published Application Once you click on the exported link, you will be taken to the application page.
### Sessions You can easily access all historical interactions of your users with your application.
### Logs You can easily access logs and all the processed information from the logs page.
### Invite You can invite team members and collaborators to work together on an application. Similarly, you can also add team members to folder from the folders page which gives them access to all the applications in the folder.
# Knowledge Base Easily upload, manage and review documents and information to the knowledge base. You can think of knowledge base as the central information system of the application. The documents uploaded are indexed and stored in a vector database for easy retrieval.
Users can upload multiple document formats to the knowledge base. 1. **Documents** - Easily upload any of these file types with one click - .pdf, .epub,.pptx, .docx, .rtf, .txt. 2. **Website** - Easily scrape information and data by providing the web url. 3. **Images** - Along with uploading the image, the loader extracts all the information visible in the image including texts, figures, objects, description. 4. **Audio** - Along with uploading audio files to the knowledge base, system transcribes the audio file and extract all the information. Allowed file types - .3ga, .aac, .ac3, .flac, .flv, .m4a, .mp3, .wav, .wma. 5. **Video** - Along with uploading the video files to the knowledge base, system transcribes the video and extract all the information. Allowed file types - .mts, .mov, .mp2, .mp4, .m4v, .webm. **More formats coming soon......** # Layout Elements Layout elements are essentially used to provide description and information to the users about the application. You can add headings, paragraphs, separators and URL links.
Once you drag the element to the workflow builder, you can easily customize the element by clicking on the element.
To remove an element, just click on the trash icon. # Input Elements Input elements gives you the flexibility to allow users to enter text / query which can be used in the workflow. You get access to multiple input types - checkbox, date, number, select (dropdown), text and textarea fields.
You can easily customize the element to provide information about the purpose of the input field.
To remove an element, just click on the trash icon. # Knowledge Base Retriever The knowledge base retriever can be used to search the knowledge base and retrieve relevant information from the vector store based on the query. You can retrieve information from documents, websites, images, audios and videos uploaded to the knowledge base. Depending on your use case, you can use single or multiple retrievers.
Once you drop the retriever to the builder zone, you need to configure the query to retrieve relevant information. You can hardcode your query text in the query box.
Alternatively, you can pipe in the variable from the workflow to be used as a query. Type "**#**" in the Query box, it will show the list of available variables from the workflow builder. Users can also specify number of results to be fetched.
**These steps are applicable to all knowledge base retrievers.** # UI Data Loaders UI data loaders give users flexibility to add real-time documents and data to the workflow.
The UI data loaders functions same as the ones available for knowledge base retrievers. The advantage of UI data loader over knowledge base is that it gives the flexibility for the frontend users to add their own document in the workflow for processing. Admin can also select if they want the documents uploaded by the users to be added to knowledge base or not for future processing. Similar to the knowledge base loaders, admin can either add static query for information retrieval or pipe variable from the workflow.
**These steps are applicable to all UI data loaders.** # Agents Agents are off-the-shelf tools which can help you enhance your workflow. These are collection of toolkits designed to do a specific task. You can also select if the output of agents are visible to the users or hidden.
### Google Search You can use Google Search agent to perform real-time search on Google and retrieve answers. You can either hard code the search query or pipe a set of variables from the workflow.
### Wikipedia You can use Wikipedia agent to perform real-time search on Wikipedia and retrieve answers. You can either hard code the search query or pipe a set of variables from the workflow.
### YouTube YouTube agent helps users to extract transcript by providing a YouTube video url. You can either hard code the url or pipe a set of variables from the workflow.
# Actions Actions are core components of the application. Actions uses large language model that takes data from input elements, knowledge base and agents as inputs and returns messages as outputs based on the prompt. You can specify system instructions, prompts, interchange between different providers and models. Each action step can either be final output or intermediate depending one the use case.
### Text to Text Creates text from text input. You need to specify the task in the prompt box. You can use the input variables as well as result generated from previous steps as input to prompt.
Model Setting - In the model setting tab, you can switch between different LLM providers, models, temperature and maximum output length.
### Summarizer Summarizer action can be used to review a document and generate a summary of the content.
*Currently, summarizer only works with UI Data Loaders.*