Globant Enterprise AI, the platform developed with GeneXus that simplifies the creation of AI agents and assistants tailored to each business's needs without requiring programming, has released over 20 enhancements that expand its flexibility and power.

The March updates include state-of-the-art models such as GPT-4.5 and Claude 3.7, RAG improvements with Pinecone support, Full Story integration for user analysis, the ability to attach files in Flows, and a more agile and efficient Data Analyst Assistant.

“We are focused on offering users the best combination of autonomy, control, and security, ensuring they can experiment, iterate, and scale their solutions seamlessly. With each update to Globant Enterprise AI, we aim not only to enhance the technology but also to transform and simplify how businesses integrate artificial intelligence into their processes, making it more accessible, efficient, and secure,” adds Eugenio García, Product Manager at Globant Enterprise AI.
Here are the most notable updates for March:
 

New Evaluation APIs: Enhancing Model Analysis

Evaluating the performance of your AI Assistants is essential to ensure they provide accurate and useful responses. The Evaluation feature offers data scientists an objective and structured method to measure the impact of changes in prompting, agent configurations, or the use of different LLMs. This allows them to precisely assess whether these modifications improve the quality of the agent's responses and continuously optimize its performance.
The Evaluation feature enables:
 

• Defining a set of test questions
  Create a list of questions specifically designed to assess your AI Assistant's performance.
• Running evaluations
  Conduct these tests on different versions of your AI Assistant, automatically collecting responses and feedback.
• Analyzing evaluation results
  The module provides dashboards and visualizations to track the evolution of your AI Assistant’s performance across different runs. You can monitor metrics such as the percentage of correct responses and identify areas for improvement.
• Automating feedback collection
  Configure the module to automatically compare the AI Assistant’s responses with expected answers and generate feedback without manual intervention.

 

New LLMs

To enhance generative AI development, we’ve added new models to the 100+ LLMs already integrated into the platform, expanding the range of options for customizing your enterprise solutions. The newly added LLMs include:

• GPT-4.5 – OpenAI’s latest release, featuring improvements in performance, speed, and contextual understanding.
• Claude 3.7 Sonnet – A more advanced model from Anthropic, now also available on Vertex AI and AWS Bedrock, allowing seamless adaptation to your tech stack without compatibility concerns.

 

Gemini 2.0 Series Enhancements

Google continues to evolve with Gemini 2.0, and Globant Enterprise AI is keeping pace by integrating two new preview versions into Vertex AI. These updates provide greater speed and precision, enabling businesses to maximize generative AI capabilities without sacrificing performance:
 

• vertex_ai/gemini-2.0-flash-lite-preview-02-05 – Optimized for fast responses and efficient processing.
• vertex_ai/gemini-2.0-flash-thinking-exp-01-21 – Designed for more complex tasks requiring structured reasoning.

 

RAG Improvements

The sixth revision of RAG brings key enhancements to optimize information retrieval and processing:

• Support for o3-mini, gpt-4.5-preview, claude-3-7-sonnet-20250219, DeepSeek, Gemini 2, and Sambanova*. You can find the full list of supported LLMs [here].
• New Pinecone provider available for embeddings and rerankers, improving the accuracy of generated responses.
• "CleanUp" action fixed – This function now correctly indicates that files will be permanently deleted from the RAG Assistant and updates information in RDS.
• New usage indicator to enhance AI consumption monitoring.
• Improved LLM and embedding configurations – All models and providers are now standardized in dropdown lists, making selection more intuitive and consistent. If you need custom options, you can use the override mechanism.
• Added support for guardrails, providing enhanced security and control over response generation.
• New "documentAggregation" property – Allows you to define how information sources are grouped and presented.

 

New Feedback System in Chat with Data Assistant

You can now provide direct feedback on responses via the frontend, improving user experience and enhancing assistant accuracy.
 

File Attachments in Flows

Now you can attach files to your Flows, improving data interaction and streamlining information processing within the platform.
 

Full Story Integration in Workspace/Playground

Support for Full Story has been added, enabling detailed user access statistics in the Workspace and Playground. This provides a more comprehensive view of platform usage, helping optimize user experience and workflows.
 

LLM API Enhancements

Now, in the LLM API, models that support multiple languages will include the "descriptions" property in API responses. This allows access to descriptions in Spanish, English, and Japanese, improving accessibility and customization when interacting with AI models.
 

Data Analyst Assistant 2.0: Simpler and More Efficient

The new Data Analyst Assistant 2.0 version introduces significant improvements to optimize data analysis:
 

• Simplified Structure – Now includes only two main components: Dispatcher and Thinker, reducing complexity in data interaction.
• Automatic Metadata Generation – When uploading datasets, the metadata structure is automatically generated, accelerating setup and reducing preparation time.



 

For more details, check out the guide on How to Use Data Analyst Assistant.

What’s Next?

In the next updates, scheduled for mid-April 2025, we will introduce:

A new AI Studio – allowing you to define, manage, and orchestrate autonomous AI agents.
 

• A new Python SDK (PyGEAI).
• A new file management module.
• And much more!

The Globant Enterprise AI roadmap is now available in the Coming Next section of the Globant Enterprise AI Documentation.