Abstract: A computer system for creating and processing forms within block-based workspaces provides connections between form questions and database properties. The computer system enables creation of forms either from scratch or based on existing database structures, with each form question mapped to a database property. The computer system includes a form builder interface for configuring questions with validation rules, a sharing interface for distributing forms publicly or within workspaces, and automation capabilities for triggering actions based on responses. Forms can be customized with various question types, styling options, and validation rules while maintaining direct connection to the underlying database. The validation rules include configurable constraints for different question types. The form responses automatically populate workspace databases, enabling immediate analysis and workflow automation.

Abstract: A system can generate and render, at a computing device, a page comprising a graphical user interface (GUI) that includes a displayable item from a first block of a block-based data structure. The system can use at least a portion of a natural language prompt received by the GUI to cause a neural network (e.g., an LLM) to generate a query executable against the block-based data structure to return a result set comprising a second block. The second block is not initially displayed at the page and can be discoverable by determining query context using at least a portion of the natural language prompt in combination with a property of the first block. The system can cause the generated query to be executed against the block-based data structure to generate and display, at the GUI, (i) an ephemeral view comprising the result set and (ii) a query tuning control configured to enable dynamic tuning of the generated query via the GUI.

Abstract: A multimodal content management system having a block-based data structure can include a query engine configured to perform automatic data discovery for natural language queries. The system can generate and render, at a computing device, a page comprising a graphical user interface (GUI) with a displayable item from a first block of a block-based data structure. The system can generate and bind, to the page, a schema definition comprising a first reference to the first block and a second reference to a set of blocks, wherein the first block is relationally linked to the set of blocks via the second reference. The system can use at least a portion of a natural language prompt, received at the GUI, to generate an input feature for a large language model, the input feature having a schema-question pair that includes at least a portion of the schema definition. The large language model can generate a query configured to operate on the block-based data structure.

Abstract: A computing system performs automated tasks within a digital environment by causing a large language model (LLM) to generate computer-readable code that is executable by the computer system to perform the tasks. The computing system causes the LLM to generate a transcript of code to perform a task within a digital environment, where the task includes a plurality of operations to be performed within the digital environment. The LLM is configured to stream the computer-readable code to the transcript as the code is generated. The computing system iteratively manipulates strings within the transcript to generate executable portions of the computer-readable code from partially streamed outputs from the LLM. The computing system sequentially executes the generated executable portions of the computer-readable code to cause a preview of each of the plurality of operations to be sequentially output to the digital environment.

Abstract: The system obtains multiple views associated with multiple messages. A view among the multiple views includes a visual appearance and a criterion among the multiple criteria defining a message to include in the view. The system obtains the multiple messages and analyzes the multiple messages using the multiple criteria to determine a number of messages satisfying the criterion. The system determines whether the number of messages satisfying the criterion among the multiple criteria satisfies a predetermined threshold. Upon determining that the number of messages satisfying the criterion among the multiple criteria satisfies the predetermined threshold, the system suggests organizing the number of messages satisfying the criterion according to the view. The system receives an approval to organize the number of messages satisfying the criterion according to the view and presents the number of messages according to the visual appearance associated with the view.

Abstract: The system obtains messages including multiple properties, such as a sender, a receiver, and content. The system creates a database to store the messages and presents the database in a user interface including a table including rows and columns. The database can be customized by including a custom column among the multiple columns. The system customizes the user interface by creating custom views of the database. The system receives an indication to create a view of the multiple messages, where the indication to create the view includes a criterion to select a subset of messages from the messages. The system creates the view including multiple layouts configured to visually provide important information associated with the subset of messages.

Abstract: The system obtains multiple views associated with a first multiplicity of messages, where a view among the multiple views includes a criterion, a visual appearance, and an action. The criterion defines a message to include in the view, and the action is configured to trigger automatically when the message is included in the view. The system applies the view to the second multiplicity of messages by: selecting a subset of messages among the second multiplicity of messages according to the criterion; presenting the subset of messages to the user according to the visual appearance associated with the view; and performing the action on the subset of messages.

Abstract: The system provides a table including multiple email messages in a user interface. The table includes multiple rows and multiple columns, where the multiple columns correspond to respective properties of the email messages and a row corresponds to one or more email messages. The system modifies the user interface by receiving an indication to modify the table with content of a particular type. The indication includes a selection of one or more rows of the multiple rows in the table wherein the selection of the one or more rows is across at least part of the multiple columns of the table. The system causes a generative artificial intelligence (AI) system to create content of the particular type and the selection of the one or more rows. The system modifies the user interface by adding the content of the particular type created by the generative AI system to the user interface.

Abstract: The system obtains multiple email messages and provides a user interface to display the multiple email messages. The multiple email messages can be configured to be presented in a table including multiple rows and multiple columns. The user interface can include a unique identification (ID) associated with the row, and a unique ID associated with the column. The system can customize the user interface by receiving an indication to create a graphical representation of the content of the multiple email messages. The indication can include obtaining a formula indicating an operation to be performed on a content identified by the row and/or the column. The system can create the graphical representation of the operation applied to the content identified by the row and/or the column and provide the graphical representation on the user interface.

Abstract: A system implements an interpreter for extensible markup language (XML) to validate XML produced by large language models (LLMs). The system uses an LLM to generate a first set of XML instructions to perform a first task in an environment communicatively coupled to the system. The system executes the first set of XML instructions and observes a result of the execution. Based on the observed result, the system either uses the LLM to generate a second set of XML instructions to perform a second task, or detects an error in the first set of XML instructions. When an error is detected, the system uses the LLM to modify the first set of XML instructions to correct the error.

Abstract: A multimodal content management system having a block-based data structure can include a neural network trained to operate on the block-based data structure, such as by recognizing units (e.g., block titles, block identifiers, block content/content types, block properties, block types, block dependencies, and/or block format) within the block-based data structure. The neural network can be trained on tokens (e.g., tokens generated from a natural language prompt) that relate to the aforementioned units. Training operations can include causing the neural network to generate: (i) a first set of response tokens based on a keyword included in training data and (ii) a second set of response tokens based on an inference made regarding tokens in the training data. Training operations can further include generating a GUI that includes the keyword, the inference and at least one navigable link to a unit in the block-based data structure.

Abstract: A multimodal content management system having a block-based data structure can include a question and answer (Q&A) assistant engine. The Q&A assistant engine can capture a natural language prompt, which can be tokenized to generate a set of data source tokens and a set of parameter tokens. A parameter token in the set of parameter tokens can be indicative of a block property in the block-based data structure. A trained neural network can accept the data source tokens and/or parameter tokens as inputs to generate a query executable against the block-based data structure. The query can generate a result set, which can be visualized concurrently with the input control of the Q&A assistant engine. The input control can accept subsequent natural language prompts while maintaining result set visualizations. Items in result set visualizations generated in response to a particular prompt can include content in different modalities.

Abstract: A multimodal content management system having a block-based data structure can include a question and answer (Q&A) assistant (e.g., a chatbot). The system can receive a natural language prompt and generate a result set. The result set can include blocks (e.g., blocks that include responsive content, including content in different modalities). The system can apply a set of authority signals to items in the result set to generate a ranked result set. The authority signals can be generated using aspects of the block-based data structure, such as block properties. The system can cause the Q&A assistant to return a set of hyperlinks to the ranked result set items. The hyperlinks can be operable to enable navigation to block content without closing the Q&A assistant.

Abstract: A multimodal content management system having a block-based data structure can include an artificial intelligence (AI)-based code unit generator that can generate code units executable against the block-based data structure to provide information requested by users. For example, the code units can be generated in response to natural language prompts received via a question and answer Q&A assistant engine. A neural network can be trained on block types, block dependencies, block content values, block content types, and/or block format. The neural network can receive a set of tokens generated based on a natural language prompt and generate one or more query strings to be included in a particular code unit. The tokens can be indicative of block properties, content, or other items in the block-based data structure. The code unit can be structured to execute more than one query against the block-based data structure such that a particular result set can include content items of different modalities.

Abstract: A multimodal content management system having a block-based data structure can include an artificial intelligence (AI)-based embeddings generator and indexer. After receiving an item update instruction that includes an object (e.g., a block content, a block property, or a block schema) identifier and an update payload, the system can transform the update payload—for example, by generating a chunk to capture at least a portion of the update payload. The chunk can correspond to a particular content modality included in the update payload. The system can generate and retrievably store a vector comprising a set of embeddings corresponding to the chunk, where the embeddings represent a vectorized portion of block content, block property, or block schema.

Abstract: Methods and devices for an automation flow in a digital workspace are disclosed. In one example aspect, a method includes providing a user interface for generating an automation associated with a database table in the digital workspace. The user interface includes a trigger control element to define a trigger for the automation, and the trigger control element comprises an option to add a button element as the trigger for the automation. The method includes populating, within the user interface, a set of selectable options for the button element that includes an indication for a relationship between the button element and the database table. The method also includes executing one or more actions of the automation upon the button element being clicked.

Abstract: A content platform displays a user interface for defining a formula to perform within a table, where the table includes a plurality of rows and a plurality of columns. An identifier of each of the plurality of columns is populated into the user interface for selection by a user. A selection is received, via the user interface, of a first identifier corresponding to a first column and an input to define an operation to be performed on a value in the first column. The formula is stored in association with a second column of the table, and can be executed to modify a value in the second column of the table based on performing the operation on the value in the first column.

Abstract: A system implements an interpreter for extensible markup language (XML) to validate XML produced by large language models (LLMs). The system uses an LLM to generate a first set of XML instructions to perform a first task in an environment communicatively coupled to the system. The system executes the first set of XML instructions and observes a result of the execution. Based on the observed result, the system either uses the LLM to generate a second set of XML instructions to perform a second task, or detects an error in the first set of XML instructions. When an error is detected, the system uses the LLM to modify the first set of XML instructions to correct the error.

Abstract: A system implements an interpreter for extensible markup language (XML) to validate XML produced by large language models (LLMs). The system uses an LLM to generate a first set of XML instructions to perform a first task in an environment communicatively coupled to the system. The system executes the first set of XML instructions and observes a result of the execution. Based on the observed result, the system either uses the LLM to generate a second set of XML instructions to perform a second task, or detects an error in the first set of XML instructions. When an error is detected, the system uses the LLM to modify the first set of XML instructions to correct the error.

Abstract: A system provides an artificial intelligence assistant to automate tasks within a structured environment. The system receives a natural language instruction to perform a task within an environment that is communicatively coupled to the system. Based on the received instruction, the system generates a computer-readable input that includes a context of the environment and a computer-readable form of the received instruction. The system sends the computer-readable input to a large language model (LLM) to cause the LLM to generate a set of computer program code to perform the task. The system executes the computer program code to perform the task in the environment.