Abstract: Certain aspects of the present disclosure provide techniques for providing smart content to a user of an application. Embodiments include receiving a request from a client for content. The request may include context data. Embodiments include identifying a content template for the content based on the request. Embodiments include identifying a rule associated with the content template. Embodiments include evaluating the rule based on the context data in order to determine a value of a variable. Embodiments include generating personalized content based on the content template and the value of the variable. Embodiments include providing the personalized content to the client.

Abstract: A method including extracting data from disparate data sources. The data includes data pairs including a corresponding data point and a corresponding time associated with the corresponding data point. The method also includes extracting insights from the data at least by identifying a trend in the data pairs. The method also includes forming a model vector including the insights and an additional attribute to the insights. The additional attribute characterizes the insights. The additional attribute includes at least user feedback including a user ranking of a ranked subset of the insights from a user. The method also includes inputting the model vector into a trained insight machine learning model to obtain a predicted ranking of the insights. The method also includes selecting, based on the predicted user ranking, a pre-determined number of insights to form predicted relevant insights. The method also includes reporting the predicted relevant insights.

Abstract: Systems and methods are disclosed for converting natural language queries to a query instruction set for searching a data warehouse. To generate a query instruction set from a natural language query, a system iteratively uses a generative artificial intelligence (AI) model and database query tools to generate a query instruction set in a stepwise manner. The system and generative AI model do not require a priori knowledge of data table contents in the data warehouse, which may include sensitive information. In addition, the system does not require access to the data warehouse to generate the query instruction set. Instead, the system is implemented to use structure information from the data warehouse, including table lists (such as table names) and table format information (such as column names) of tables in the data warehouse, and the generative AI model is a generally trained model to generate the query instruction set.

Abstract: A method including receiving, at a large language model, a prompt injection cyberattack. The method also includes executing the large language model. The large language model takes, as input, the prompt injection cyberattack and generates a first output. The method also includes receiving, by a guardian controller, the first output of the large language model. The guardian controller includes a machine learning model and a security application. The method also includes determining a probability that the first output of the large language model is poisoned by the prompt injection cyberattack. The method also includes determining whether the probability satisfies a threshold. The method also includes enforcing, by the guardian controller and responsive to the probability satisfying the threshold, a security scheme on use of the first output of the large language model by a control application. Enforcing the security scheme mitigates the prompt injection cyberattack.

Abstract: Systems and methods are provided for using vector embeddings and large language models to answer chatbot inquiries.

Abstract: A method includes receiving, at a server from a user device, a user query to a large language model (LLM), creating an LLM query from the user query and an application context, gathering confidential information from the LLM query, and sending the LLM query to the LLM. The method includes receiving, from the LLM, an LLM response to the LLM query, comparing the LLM response to the confidential information to generate comparison result, and setting a leakage detection signal based on comparison result.

Abstract: A method includes receiving, at a server from a user device, a user query to a large language model (LLM), creating an LLM query from the user query, inserting a system prohibited request into the LLM query to generate a revised LLM query, and sending the revised LLM query to the LLM. The method further includes receiving, from the LLM, a first LLM response to the LLM query, testing the first LLM response to detect whether a prohibited response to the system prohibited request is included in the first LLM response, and setting a prompt injection signal based on whether the prohibited response to the system prohibited request is included in the first LLM response.

Abstract: A method includes receiving, at a server from a user device, a user query to a large language model (LLM), creating an LLM query from the user query, inserting an security marker instruction into the LLM query to trigger an injection of a security marker, and sending the LLM query to the LLM. The method further includes receiving, from the LLM, an LLM response to the LLM query, evaluating the LLM response to detect whether the security marker is present in the LLM response, and setting a prompt injection signal based on whether the security marker is present in the LLM response.

Abstract: Systems and methods are disclosed for managing categorization problem solutions and identifying miscategorizations. The identification of a miscategorization of an object is based on the object's first embedding being different than the first embeddings of other objects in a cluster. The objects in the cluster are clustered together based on second embeddings of the objects, with the first embedding generated based on a first description associated with an object and the second embedding generated based on a second description associated with the object. As such, while the clustering of second embeddings may initially indicate that the objects in the cluster are similar, the comparison between first embeddings of the objects in the cluster (such as calculating a distance between a first embedding and a center of the cluster based on the first embeddings) can confirm whether an object in the cluster is different and thus is potentially miscategorized.

Abstract: Systems and methods for determining ownership of cloud computing resources are disclosed. An example method includes identifying a first active table whose ownership is not defined in a central repository, determining, based on a write log associated with the first active table, a first timestamp and a first internet address associated with a most recent write to the first active table, determining, based on the first internet address, whether or not the first timestamp is more recent than a creation time of a first cloud computing instance corresponding to the most recent write, and in response to the first timestamp being more recent than the creation time of the first cloud computing instance, identifying a first owner of the first active table based on a first cost allocation tag associated with the first cloud computing instance.

Abstract: A method that includes extracting image features of a document image, executing an optical character recognition (OCR) engine on the document image to obtain OCR output, and extracting OCR features from the OCR output. The method further includes executing an anomaly detection model using features including the OCR features and the image features to generate anomaly score, and presenting anomaly score.

Abstract: A computer-implemented method including tracking data describing a new population of users of a software application including different graphical user interfaces (GUIs). The method also includes generating a distribution by mapping the data to lookalike cohorts. The method also includes extracting, using a random sampling algorithm, samples from the distribution. The method also includes generating, from the samples, a normal distribution of predicted long term values of the new population of users. The method also includes selecting an expected long term value from the normal distribution. The method also includes generating, from the normal distribution, an estimated distribution, around the expected long term value, of estimated long-term values for the new population. The method also includes selecting, using the expected long term value and the estimated distribution, a selected GUI from among the different GUIs. The method also includes modifying the software application by presenting the selected GUI.

Abstract: A processor may filter data to generate a subset of the data less than an entire set of the data. The subset may include at least one string and at least one numeric value. The processor may match the at least one string and the at least one numeric value to one of a plurality of archetypes by applying a clustering algorithm. Each archetype may include a subset of archetype data less than an entire set of archetype data. The processor may compare the entire set of data to the entire set of archetype data to identify at least one difference between the entire set of data and the entire set of archetype data. The processor may apply at least one optimization to address the at least one difference.

Abstract: Systems and methods for enriching raw user text with a database to identify relevant context, wherein generated prompts provide responses to user queries is provided. A method includes receiving a query, wherein the query comprises the raw text, creating a first embedding based on the query, retrieving a plurality of other embeddings, wherein the plurality of other embeddings are complementary to the first embedding, creating a contextual prompt including context from at least one of the plurality of other embeddings, processing the contextual prompt using a trained machine learning model, thereby generating a response to the query, and causing the response to be displayed by a display device.

Abstract: A computing system identifies an incoming voice call from a user device to an agent device associated with the computing system. The computing system generates a transcription of the incoming voice call using one or more natural language processing techniques. The computing system extracts a problem description from the transcription. The problem description indicates a topic for the incoming voice call. A first machine learning model estimates a situation vector from the problem description. A second machine learning model identifies a pre-existing situation vector that closely matches the estimated situation vector. The computing system retrieves a situation description that corresponds to the identified pre-existing situation vector.

Abstract: A machine learning system executed by a processor may generate predictions for a variety of natural language processing (NLP) tasks. The machine learning system may include a single deployment implementing a parameter efficient transfer learning architecture. The machine learning system may use adapter layers to dynamically modify a base model to generate a plurality of fine-tuned models. Each fine-tuned model may generate predictions for a specific NLP task. By transferring knowledge from the base model to each fine-tuned model, the ML system achieves a significant reduction in the number of tunable parameters required to generate a fine-tuned NLP model and decreases the fine-tuned model artifact size. Additionally, the ML system reduces training times for fine-tuned NLP models, promotes transfer learning across NLP tasks with lower labeled data volumes, and enables easier and more computationally efficient deployments for multi-task NLP.

Abstract: A method including detecting, in a written electronic communication, an input sentence satisfying a readability metric threshold. The method also includes transforming, by a sentence transformer model, the input sentence to output suggested sentences. The method also includes evaluating the suggested sentences along a set of acceptability criteria. The method also includes determining, based on the evaluating, that the set of acceptability criteria is satisfied. The method also includes modifying, based on determining that the set of acceptability criteria is satisfied, the written electronic communication with the suggested sentences to obtain a modified written electronic communication. The method also includes returning the modified written electronic communication.

Abstract: A method including receiving an input including a number of texts from a source of text and a number of images from a source of images. The texts are separate from the images. The input is embedded into a first data structure that defines first relationships among the images from the source of images and the texts from the source of text. The first data structure is compared to an index including a second data structure that defines second relationships among a number of pre-determined texts and a number of pre-determined images. The pre-determined texts have known relationships to the pre-determined images. Each pre-determined image in the pre-determined images is related to one or more instances of the pre-determined texts. A subset of images, those images in the pre-determined images for which matches exist between the first relationships and the second relationships, is returned from the pre-determined images.