Skip to main content

AI Assistant

Capabilities​

Our AI Assistant, powered by OpenAI's GPT-4 language model, provides an advanced and interactive approach to data analysis and visualization. This feature leverages natural language processing (NLP) to interpret text prompts and automate various tasks within the app. These tasks include:

Voice commands​

You can use voice commands to interact with the AI Assistant. Click on the microphone icon in the prompt input area to start speaking. The AI Assistant will transcribe your speech and provide a response based on your query.

AI Assistant

Prompt history​

The AI Assistant keeps a personal searchable history of useful prompts that you like & save. This is particularly useful when you want to send the same or similar prompts in the future.

To save a prompt to your personal history, click on the save icon above the prompt. To view your saved prompts, type '@' in the prompt input area.

To search your prompt history, type '@' followed by a search term. For example, '@bar' will show all saved prompts containing the word 'bar'.

AI Assistant

Chat history​

The AI Assistant keeps a history of all your chats. You can scroll through the chat history to review previous interactions with the Assistant. You can also resume a previous conversation by clicking on the chat history entry, and delete a chat history entry by clicking on the delete icon.

Related files. Chats are automatically associated with the files you are working on, so you can directly open related files from the chat history panel. In the near future, you will be able to search for chats related to a specific file.

PDF analysis and information extraction​

Location: My Files area

A powerful new feature allows users to upload, select, and query article PDF documents in My Files area. This capability utilizes state-of-the-art Large Language models to extract relevant information from the PDFs, providing users with quick and insightful answers to their questions.

Research articles as primary focus

This feature currently focuses on research article PDFs, processing of other types of PDFs may be less efficient.

Uploading and managing PDF files​

  • Go to the My Files area in the platform.
  • Click on the Upload button to select and upload one or multiple PDF files. The files will be securely stored in your personal space.
  • Once uploaded, the PDF files can be viewed and managed in the My Files area. Users can rename, delete, or select files for analysis.
  • To analyze a PDF, simply click on it in the My Files area. You can select multiple PDFs for a broader analysis.
  • Selected files will be highlighted, indicating they are ready for querying.
  • With the PDFs selected, you can ask the AI Assistant various questions about the content of these documents. The Assistant will process your query and extract relevant information from the selected PDFs.

How it works​

The technology behind this feature involves converting the contents of the PDFs into embeddings, a form of data representation that captures the essence of the text. When a query is made, a semantic search is performed within the embedding database to find and extract the most relevant information.

Examples of queries​

  • Summarize the key findings in the selected PDFs.
  • What experimental models were used?
  • What was the experimental approach?

Limitations and best practices​

  • Quality of PDFs: The accuracy of information extraction depends on the quality and clarity of the PDF documents.
  • Complex Queries: While the AI can handle a wide range of queries, overly complex or ambiguous questions may result in less precise answers.
  • Language Limitations: Currently, the feature supports PDFs primarily in English. Additional language support is being developed.

Spreadsheet calculations and data generation​

Location: Data & Visualization area

The AI Assistant can perform a variety of calculations in the data grid and add new data to it based on your prompts. This is particularly handy when you need to quickly perform operations on existing data or add computed values to your data grid.

Examples of prompts to perform calculations or add data to the grid​

  • Calculate row-average of columns A-C, put in column D.
  • Add gaussian-distributed random data to column A with mean 0 and standard deviation 1.
  • Calculate average of column A, put in B1.

Advanced data analysis​

Location: Data & Visualization area

The Assistant can perform advanced data analysis using the latest GPT-4o model from OpenAI and its code writing capabilities. You can ask the Assistant to clean, interpret and analyze entire datasets or specific columns of data. The Assistant can also provide insights into the distribution, correlation, and other statistical properties of the data, including hypothesis testing, as well as generate new or derived datasets. This feature can be seamlessly integrated with the data visualization area to create plots based on the analyzed data.

Note on data privacy

The data you want to analyze and column headers will be sent to OpenAI! Read more about OpenAI's data policies.

Examples of data analysis prompts​

  • Please clean up this dataset
  • Tell me what you think about data in columns A-B (they are linked).
  • Are columns A and B normally distributed?
  • What is the correlation between columns A and B?

Data visualization​

Location: Data & Visualization area

Traces​

Figlinq's plots are based on plotly.js, the open-source JavaScript graphing library. Plotly.js plots are composed of traces. The AI Assistant can create a variety of trace types, including bar, scatter, line, pie, heatmap, and many more.

With the AI Assistant, you can easily create a new trace by simply typing in a text prompt. The Assistant will interpret your prompt and generate a trace based on the data from specified data grid columns. You can also specify the type of trace you want to create, or data sources for various trace properties.

Adding traces only in data & visualization area

Adding new traces is not possible in figure editor. This is because the AI Assistant needs access to the underlying data grid.

Examples of prompts to add a new trace​

  • Create a bar trace using column A for x and column B for y.
  • Create a scatter trace from columns A and B. Use column C for marker color and column D for size.
  • Create a heatmap from columns A and B. Use columns C-H for z. Axis titles: x-axis: 'axis x', y-axis: 'axis y'.

Creating figures​

Location: Figure editor area

The AI Assistant can also modify existing plots based on your prompts. You can specify the type of modification you want to make, the trace you want to modify, and the data sources for the modification. The Assistant will interpret your prompt and modify the plot accordingly. This feature is available in both the data visualization area and the figure editor.

Examples of prompts to modify a trace or plot​

  • The first trace should use column A for x and column D for y. It should also use the rithe y-axis.
  • Set tick spacing for the x-axis to 0.5 and for the y-axis to 0.1
  • Set the title of the plot to 'Welcome AI'.
  • Increase the plot width by 25% and the plot height by 50%.

Writing and editing text​

Location: Text editor area

The AI Assistant is a powerful tool for creating and refining text documents. Whether you're drafting a report, writing an article, or editing an existing document, the Assistant can help streamline your workflow with natural language commands.

Creating and editing documents​

You can use the AI Assistant to draft new documents from scratch or edit existing text. The Assistant can assist with a variety of tasks, including grammar and style corrections, content expansion, summarization, and more.

Examples of prompts for writing and editing text

  • Write an introduction about climate change.
  • Summarize this document.
  • Expand on the second paragraph, providing more details.
  • Correct the grammar and style of this section.

Writing articles and reports using STORM.​

Location: Text editor area

From the STORM GitHub repository:

STORM is a Large Language Model system that writes Wikipedia-like articles from scratch based on Internet search. (...)

How STORM works STORM breaks down generating long articles with citations into two steps:

  • Pre-writing stage: The system conducts Internet-based research to collect references and generates an outline.
  • Writing stage: The system uses the outline and references to generate the full-length article with citations. STORM identifies the core of automating the research process as automatically coming up with good questions to ask. Directly prompting the language model to ask questions does not work well. To improve the depth and breadth of the questions, STORM adopts two strategies:

Perspective-Guided Question Asking: Given the input topic, STORM discovers different perspectives by surveying existing articles from similar topics and uses them to control the question-asking process. Simulated Conversation: STORM simulates a conversation between a Wikipedia writer and a topic expert grounded in Internet sources to enable the language model to update its understanding of the topic and ask follow-up questions.

To use STORM, you can provide a topic or a question to the AI Assistant, and it will generate a Wikipedia-like article based on the topic or question. The Assistant will conduct Internet-based research to collect references, generate an outline, and the full-length article with citations.

Search Mode​

Location: Everywhere

Search mode in our AI Assistant leverages the Perplexity to provide comprehensive search capabilities. This advanced feature allows users to search through a vast range of documents and datasets, extracting the most relevant information using sophisticated natural language processing algorithms. Search mode utilizes the Perplexity service to interpret and execute search queries. When a query is made, the API processes the request, performs a semantic search across the indexed documents or datasets, and returns the most relevant results in the form of a chat reply.

Enabling Search Mode​

To enable search mode, just flip the switch above the prompt input area.

AI Assistant

Examples of Queries​

Here are some examples of how you can utilize search mode with the AI Assistant:

  • Find recent studies on climate change impacts.
  • What are the latest trends in AI research?
  • Summarize the key findings of the top three articles on renewable energy.