What is ExpertAI vs. what it is not, and differentiators
How ExpertAI is different from external AI tools
Technology differentiators
Document segmentation and vector database
ExpertAI breaks down documents into smaller parts and stores the relevant information together in a vector database. This allows the system to use Expert knowledge as the foundational source of truth for a customer's information, which ensures that answers are derived only from the content stored in the customer's knowledge base.
Generative AI with guardrails
When a user asks a question, ExpertAI retrieves the relevant Kernels (smaller parts of articles) from the vector database. This information is then sent to a large language model (LLM), which converts it into a conversational AI response.
The critical difference here is that ExpertAI will not answer a question if it does not have enough relevant or confident information, ensuring accuracy and reliability.
Integration capabilities
API Integration through an ExpertAI endpoint offers access to Expert Knowledge Kernels. This endpoint provides comprehensive information about the questions asked, the knowledge Kernels returned, and the confidence scores assigned to each. Additionally, each Kernel includes the URI of the source knowledge article.
This integration allows customers to take full control over the completion process. They can decide which language model to use for generating responses, and determine how and where the information is displayed. This flexibility ensures seamless integration of ExpertAI into existing systems and workflows, allowing organizations to maintain control and customization over their KM solutions.
The API endpoint is: GET {site-url}/@api/deki/llm/completion/report?month=YYYY-MM
Key differentiators from external tools
Source of truth
Google's Bard, OpenAI's ChatGPT, or Microsoft's Bing AI rely on publicly trained datasets, while ExpertAI exclusively uses a company's knowledge content to provide informed answers. This approach ensures that the answers are always faithful to the customer's internal knowledge base.
While asking an external LLM about a company might yield information from competitor sites, unfavorable reviews, or other sources the company cannot control, ExpertAI answers are generated solely from the customer's KB content so they control the information and how it is used.
Confidence and relevance
ExpertAI only provides answers when it has enough relevant information and confidence in the response. Otherwise, it will not attempt to answer the question.
Unlike other tools that might generate speculative responses, ExpertAI will only generate responses that can be supported by content in the customer's knowledge base.
User experience and value proposition
Comprehensive platform
Expert is a comprehensive platform where knowledge can be stored, maintained, consumed, and updated within one interface. This end-to-end solution simplifies knowledge management by providing a seamless and integrated user experience. It ensures that the knowledge is easily accessible and up-to-date, optimizing the overall workflow and efficiency.
Interaction model
While users of external LLM tools might be accustomed to a back-and-forth conversational experience, ExpertAI treats each question as a single interaction without persistence from one question to the next.
Security and integrity
The combined approach of a comprehensive platform and ExpertAI's interaction model helps prevent prompt injection and ensures that guardrails remain intact, which safeguards against manipulation and ensures that the AI remains faithful to the knowledge content, providing a secure and reliable user experience for improved product adoption, product usage, and customer satisfaction.
Specific use cases
ExpertAI excels in scenarios where the accuracy and reliability of information are paramount. It is particularly beneficial in environments where:
- Precise and context-specific answers are required.
- Code examples
- Product type / version control
- There is a need to maintain strict adherence to internal knowledge content.
- Regulatory and compliance documents
- Code of conduct and SOPs
- Publicly trained AI tools cannot be relied upon due to the sensitivity of information.
- Proprietary or confidential documentation
Internal agent use case
ExpertAI ensures the consistent provision of accurate and context-specific answers based on the company's internal knowledge base. ExpertAI performs as well as your best agents, and elevates all agents to be as good as your best agents. By leveraging ExpertAI, organizations can ensure consistent high-quality responses, enhancing the overall performance of their teams and reducing training time for new agents.
External public content use case
In public-facing scenarios, ExpertAI guarantees that sensitive and accurate information is provided without relying on publicly trained AI tools. This is crucial for environments where the sensitivity of information is high, such as in customer service or public communications. ExpertAI’s strict adherence to the company's internal knowledge base ensures that all responses are accurate and reliable, safeguarding the integrity of public communications.