We've been busy these past few months! I'd like to share some highlights of our recent product updates:
Semantic DocPrep
We rolled out our new Semantic DocPrep service to address a critical challenge in generative AI: hallucinations.
When LLMs are fed unstructured or complex data, they can sometimes generate inaccurate or misleading information. Our API service is designed to prevent that by transforming complicated documents into a structured, semantic format. Think, long form PDFs like invoices, or financial statements involving a mix of graphs, tables and text.
By providing a clean and organized input, we ensure the AI receives the most accurate information, leading to more reliable and truthful outputs. And with reliable outputs, the possibilities for GenAI to enhance your business are endless!
Autonomous Agent Builder
Next, we launched our Autonomous Agent Builder. This tool enables the quick creation of smart AI agents that can handle a wide range of complex, multi-step tasks. Our autonomous agent builder is revolutionizing processes like customer support, market research, and insurance claims processing by making them more efficient. What sets our builder apart is the extensive library of cutting-edge tools available to these agents. An agent is only as capable as its tools, and we've equipped ours with a diverse and powerful set of features:
- Plan and update a plan: Just like a human teammate, an AI agent, when given a task, can make a plan and continually update it based on what it has learned.
- Spreadsheet analysis and manipulation: This allows the agent to analyze data in spreadsheets and make necessary updates.
- Analyze images: Our agents can detect elements in images that a human might miss, such as handwritten signatures on scanned PDF documents.
- Document generation and analysis: A core function of business agents is to analyze vast amounts of data, including documents, and generate summaries of findings or suggestions.
You can read more about all the tools available here.
Additional Product Releases
We have also rolled out several other impactful enhancements:
Experience
- Content Upload & Collection Management: We've optimized the upload experience to be more user-friendly with new ways to upload files and folders of content. Vertesia can now detect changes to existing content and prevent duplicate uploads. We've also added collections as a way to organize and manage content for both users and AI agents.
- Agentic Workflow & AI Agent Observability: We've released new workflow views, offering unparalleled observability over agentic workflows and long-running conversations with autonomous AI agents. This sets a new bar for enterprises operating GenAI apps and agents in production.
Advancing AI
- New Image Generation & Video Understanding Models:
- Added support for image generation and editing using Google Imagen 4 models. This enables flexibility to generate images using multiple providers and models.
- Support for Amazon Nova video understanding. This enables video as an input modality and alternative to processing videos using speech-to-text.
- Model Tools: Added model support for tools. This allows models from Anthropic, OpenAI, Google, and more, to leverage external tools during generation, expanding their capabilities and use in agents that support tools or function calling.
Enterprise Ready
- Object Versioning: Objects in Vertesia enhance the knowledge and memory available to GenAI apps and agents. We've added object versioning, providing the ability to edit properties and compare versions. This improves tracking and managing changes made by users and AI agents.
- Single Sign-On: We’ve added enterprise single sign on for any OIDC provider (for example, Microsoft Entra and Okta).
To wrap up...
We are continuously adding new product features and capabilities to the Vertesia platform. Be sure to follow us on LinkedIn and subscribe to our blog to hear about the latest updates as they are announced.