Imagine having a 24/7 digital assistant that can handle customer service inquiries, provide technical support, offer training and education, act as an information concierge, and even assist with scheduling. These digital assistants are designed to understand and respond to user inputs with natural facial expressions, gestures, and support for over 70 languages, making interactions more human-like and effective.

Dell Technologies are now offering the Dell Generative AI Solution for Digital Assistants. Announced at Dell Technologies World this week, this solution is an excellent example of the Dell AI Factory with NVIDIA in action, leveraging Dell expertise across infrastructure, software and professional services.

I’m not going to cover or indeed repeat this announcement as you can read about the digital assistant announcement here – https://www.dell.com/en-us/blog/elevate-the-self-service-experience-for-your-customers/

and (as importantly!) read the Design Guide here – https://infohub.delltechnologies.com/en-us/t/design-guide-generative-ai-digital-assistants-in-the-enterprise/

so what are we going to talk about ? well lets provide a brief overview of how this all works / comes together and what parts have Dell Technologies made easier for customers ?

Concept

Dell’s digital assistants leverage various homegrown and third-party software component and tools to create realistic and interactive personas for business applications. These assistants integrate advanced AI to support multiple functionalities including voice, text, and visual interactions, enhancing user experience and operational efficiencies in various sectors such as retail and healthcare. But there a lot of work that happens behind the scene to make all this happen

Digital Assistant Rendering Platform

The digital assistant rendering platform, also known as the digital human platform, serves as the user-facing frontend of the overall solution, providing avatar, voice, and chat capabilities. This platform is built in collaboration with UneeQ, which specializes in creating lifelike digital human interfaces. The platform offers an immersive experience through the avatar with voice and additional text chat options, allowing users to enable or disable different modalities based on their preferences or network bandwidth limitations. It interacts with the orchestration and conversation management component using speech-to-text (STT) to submit user requests and text-to-speech (TTS) to deliver responses back to the user.

Orchestration and Conversation Management – Dell Technologies Middleware and Integration

At the heart of Dell’s solution is its middleware, which ties together various components like the digital assistant rendering platform, orchestration and conversation management, and information retrieval systems. This middleware ensures seamless integration, efficient data handling, and smooth operation across all subsystems.

Key components include:

• Speech to Text (STT): Converts spoken language into written text, commonly used in voice recognition systems and transcription services.
• Text to Speech (TTS): Converts written text into spoken words, ensuring proper pronunciation by accepting complete sentences.
• Natural Language Processing (NLP): Facilitates the understanding, interpretation, and generation of human language, performing tasks such as language understanding, translation, and sentiment analysis. Our solution leverages NVIDIA’s AI Enterprise for NLP tasks.
• Prompt Engineering: Designs and optimizes prompts to guide AI model responses effectively.
• Retrieval Augmented Generation (RAG): Enhances the accuracy and richness of AI responses by retrieving relevant documents from a database. This is powered by the Pryon Retrieval Engine.
• Translation: Translates text from one language to another, crucial for multinational environments.
• Integration Hub: Connects various systems of record, enabling seamless data sharing and communication across disparate systems.

we’ll be providing more details on this in an upcoming blog so stay tuned!

Information Retrieval

This is an Advanced RAG system at its hart so information retrieval is key. Information retrieval (IR) systems manage, process, and retrieve information from extensive data sources, ensuring the accuracy, relevance, and credibility of the information retrieved by minimizing hallucination—generation of ungrounded information. IR systems, provided by Pryon, employ sophisticated algorithms to index and search data, offering quick and easy access to the most relevant information that ultimately goes back into the prompt (as the context) to the LLM

Data Sources

Data sources for IR systems include:

• Web Pages: Online documents forming the core body of knowledge.
• Product Manuals: Detailed usage, safety, and troubleshooting information.
• Knowledge Articles and FAQs: Answers to common questions and in-depth topic information.
• Rich Media, Pictures, and Videos: Visual and multimedia content for tutorials and demonstrations.
• Drivers and Downloads: Essential software for device functionality.

These data sources form the foundation for the digital assistant’s knowledge base, enhancing its ability to respond accurately and contextually to user queries. Lean more about Pryon’s IR system here – https://www.pryon.com/product/retrieval-engine

Systems of Record

Systems of record are backend systems connected to Pyron’s software that store your critical data and establish context based on user information.

Through prebuilt connectors, Pryon’s Ingestion Engine can ingest content from dozens of enterprise content repositories, including SharePoint, Box, Amazon S3, Confluence, Google Drive, and Salesforce knowledge articles. Need more? The universal connector framework enables you to rapidly develop bespoke connectors for both unstructured and structured data sources based on your needs.

Large Language Models (LLMs)

The digital assistant solution uses Meta’s Llama 3 model, following a RAG approach to improve accuracy.

The LLMs are integrated with NVIDIA’s TensorRT-LLM and hosted on NVIDIA Triton Inference Server for optimized performance. Key aspects include:

• Foundation Models: Pretrained models like ChatGPT and Llama 3 providing a starting point for customization.

Pre-Validated and Pre-Tested Solutions

Dell’s generative AI digital assistant solutions come pre-validated and pre-tested, significantly reducing the time and effort required to design, plan, and test AI systems. This ensures that businesses can quickly deploy AI solutions with confidence, avoiding the common pitfalls associated with custom development.

Comprehensive Partner Ecosystem

• UneeQ for creating lifelike digital human interfaces.
• NVIDIA for advanced AI and machine learning frameworks.
• Pryon for robust information retrieval systems.

This collaborative approach ensures that each component of the solution is optimized and seamlessly integrated, offering businesses a robust and reliable AI infrastructure.

Scalable and Modular Architecture

Dell’s generative AI solutions are designed to be scalable and modular. This means businesses can start small and scale up as needed, adding more capabilities and handling increased workloads without significant overhauls. The modular nature allows for easy customization to meet specific business needs, making the solution flexible and future-proof.

Build vs. Buy: The Dell Technologies Advantage

Building from Scratch:

• High Initial Costs: Significant investment in time, resources, and expertise.
• Complex Integration: Ensuring compatibility between various components can be challenging.
• Maintenance and Updates: Ongoing costs and effort to keep the system updated and secure.

Buying Dell’s Solution:

• Rapid Deployment: Pre-validated and tested solutions enable quick setup and go-live.
• Reduced Risk: Proven solutions reduce the risk of implementation failures.
• Expert Support: Access to Dell’s professional services for strategy, implementation, and optimization.
• Cost-Effective: Lower total cost of ownership with scalable, modular infrastructure.

Professional Services and Ongoing Support

Dell offers extensive professional services to guide businesses through every stage of their generative AI journey. From initial strategy and planning to implementation, adoption, and scaling, Dell’s experts provide the support needed to maximize the value of AI investments.

learn more here – https://www.delltechnologies.com/asset/en-us/services/managed-services/briefs-summaries/digital-assistant-service-overview.pdf

Conclusion

By choosing Dell’s generative AI digital assistant solutions, businesses can bypass the complexities and high costs associated with building AI systems from scratch. Instead, they can leverage Dell Technologies comprehensive, integrated, and scalable solutions to quickly realize the benefits of AI, drive innovation, and stay competitive in the digital age.