H2O.ai generative and predictive AI now validated on Dell AI Factory with NVIDIA
On-premise and airgapped purpose-built AI applications for Dell customers
H2O.ai is validated and pre-integrated into Dell AI Factory for ease of scale and deployment. Accelerate time to market for use case-driven AI applications with Multimodal RAG.
Leverage private, protected data to fine-tune or retrieve-augment-generate (RAG) domain-specific LLMs for multi-agent intelligent systems.
H2O.ai running on Dell AI Factory with NVIDIA
Industry's first end-to-end enterprise AI solution to converge generative and predictive AI
On-premise and airgapped purpose-built AI applications for Dell customers
Agentic AI
use private, protected data to fine-tune or apply retrieval-augmented generation (RAG) for domain-specific LLMs in multi-agent intelligent systems
Reduce TCO
with choice of on-premise or hybrid deployment of applications with DELL AI Factory
Accelerate time to production
with pre-integrated technologies
End-to-end AI
solution
with use case development from exploration to inferencing
H2O.ai ON DELL AI FACTORY USE CASE HIGHLIGHTS
ProcurementGPT
BUSINESS PROBLEMS
One of the critical challenges in procurement is ensuring that contracts exchanged between organizations, such as banks and suppliers, are aligned and free from discrepancies.
NEEDS
Need efficient, reliable tools to review contracts, manage supplier relationships, and maintain regulatory compliance, where human oversight may overlook critical contractual clauses.
- Parse documents
- Identify sections
- Break down + identify requirements
- Requirements met
- Symmetry
- Select sections
Parse documents into sections or clauses
H2O Diff Tool automatically parses documents into their respective sections or clauses as they are loaded into the app.
Identify most similar sections
Retrieve the most similar sections, including their context, from the counterpart document. This works symmetrically, scanning both left-to-right and right-to-left.
Break down a clause into requirements
Each clause may contain multiple requirements that must be fulfilled for the clause to be satisfied. These requirements can be met in different parts of the contract. We break down each clause into its key requirements to determine where each one is fulfilled.
Identify most similar requirements
Similar to finding the most similar sections, we can retrieve the most similar requirements from the counterpart document to verify whether a requirement has been met or not.
Find out which requirements were met
A green tick to the left of a requirement indicates it was met; a red cross shows it wasn’t.
The system uses the top 10 most similar requirements to determine if a requirement was met, displayed when the requirement is clicked.
An overlay highlighting met requirements in green and unmet ones in red can be activated to help easily navigate to unmet requirements.
Symmetry
The functionality is symmetrical—what works for the left-to-right document comparison also works for right-to-left.
This allows users to retrieve similar sections and requirements from either document as the base.
Manually select sections
If a section wasn’t correctly parsed or broken down into requirements, the user can manually select sections from both documents.
The user first selects a section from the template document, then from the supplier document.
Both sections are then broken down into requirements, and the system checks if the template's requirements were met by the supplier's section.
Complaint Summarizer
BUSINESS PROBLEMS
Manual processing and analysis of customer complaints can lead to missed insights and delayed resolutions. A complaint summarizer addresses these challenges by automating sentiment analysis, identifying trends, distilling recurring themes, and prioritizing urgent issues with timely feedback.
NEEDS
Faster grievance resolution and effective regulatory response. Key metrics include faster resolution times, improved first-contact resolution, and accurate complaint classification and sentiment analysis. Benefits include reduced complaint volume, higher productivity, and cost savings through automation.
- RAG workflow
- Complaints by state
- Summarizer + topic extractor
- Summary + recommendation
- LLM Agent
RAG workflow
RAG (Retrieval-Augmented Generation) workflow for summarizing customer complaints across various categories and products.
Complaint distribution by state
Displays the breakdown of complaints by state, highlighting the most frequent issues like credit cards, debt collection, and loan services.
Summary of complaints and key issues
Provides a high-level summary of the most common complaints, the top issues faced, and recommended actions to address these concerns.
Actionable insights from complaint summaries
Summarizes key topics from complaints and proposes actionable recommendations for improving services.
End-to-end LLM-powered complaint agent
Showcases the comprehensive AI-driven process from voice-to-text conversion, complaint decision-making, and complaint summarization with proposed actionable recommendations for improving services.
Enterprises are demanding AI solutions that are trustworthy, cost effective and purpose-built for domain specific use cases, and with H2O.ai, we’re delivering on that demand. The Dell AI Factory with NVIDIA and H2O.ai helps organizations deploy AI on-premises with unmatched control and cost efficiency.”
Chad Dunn
Vice President, Product Management, AI and Data Management
Dell Technologies
