Finally a solution that will take you from unstructured content to a sound decision.
Discover
How can you compare tasks, ETFs, or even laptops with precision? By leveraging semantic similarity, we map attributes like proficiency and performance to a shared vector space for meaningful comparisons. Explore how this approach enables smarter decisions across diverse domains. 🚀
Today, we’ll explore a recent enhancement to the Content Analyzer: audio file analysis. This new capability allows businesses to uncover powerful insights from voice interactions by transcribing and analyzing audio files. In this blog post, we’ll showcase how the Content Analyzer works with a real-world example—a contact center call recording. Imagine a scenario where a […]
In this blog post, we explore how to use Solver, an optimization tool, to select the best ETF based on multiple factors. By inputting criteria such as Management Expense Ratio (MER), asset value, historical returns, and sector exposure, Solver helps investors make data-driven decisions. The post walks through a practical example where we balance cost, performance, and exposure to select an ETF that meets specific investment goals. With this approach, investors can efficiently navigate the complex process of ETF selection to achieve their financial objectives.
A multi-task planning approach to tackle high-level user inquiries was developed, ensuring comprehensive and contextually accurate responses. By breaking down complex questions into detailed tasks, deriving specific questions, and thoroughly querying reference materials, we provide users with the precise information they need. Discover how this new method enhances the quality of answers, exemplified by an inquiry about credit card benefits for mobile phone damage.
These updates now support full JSON object definitions, allowing for more complex and structured data representation. Additionally, our enhanced inference logic ensures more accurate extraction of intricate data structures. Whether it’s extracting detailed flight information from a boarding pass or capturing complex data points from unstructured content, these enhancements provide unparalleled accuracy and efficiency in data extraction.
As I continue to explore what’s needed to build a robust LLM based planning agent, one thing I’ve been focusing on lately was how to improve the quality of the plan itself. One dimension I’ve been exploring is how refining the objective of the job and tasks themselves to provide a clearer and more adjusted […]
While working on research and analysis capabilities for my personal project, I started to dive into the possibility of automating further the generation of what needs to be done in order to reach a particular objective.
The Content Analyzer now offers advanced features to streamline content analysis, including question answering, PDF content ingestion, granular analysis criteria, data extraction, and proposal generation. It uses natural language processing to answer queries, extract data from PDFs, and generate actionable recommendations. These enhancements make it a powerful tool for analyzing unstructured data and driving informed decision-making.
Discusses recent improvements to the assertion API and introduces the Content Analyzer, explaining its role in content analysis. Two use cases are presented: analyzing a user story and reviewing a television. The analysis results are summarized, providing insights and potential improvements for the reviewed content.
While working on the next evolution of how content is processed on my personal project with Dretza, I had to design and build new APIs to assist in that area. These APIs offer powerful capabilities for generating assertions, determining critical aspects for content reviews and automate the content analysis process itself. There was also a […]
