Nearly a year ago, we covered Salesforce’s launch of the Einstein Trust Layer, a set of safeguards built into their Service Cloud infrastructure designed to mitigate the risks of implementing Generative AI (GenAI) in the enterprise. Since then, GenAI has continued to spark both excitement and concern as enterprise executives navigate its best uses to enhance customer and employee experiences.

With the introduction of the Einstein Trust Layer, Salesforce tackled some of the most pressing concerns about GenAI adoption: reducing the risk of unwanted behaviors from large language models and safeguarding customer and corporate privacy.

The latest challenge facing GenAI adoption is related to data. As everyone seems to agree, the ability of GenAI to provide accurate responses and effective assistance hinges on its access to comprehensive and relevant knowledge. This necessity has spotlighted the many challenges and complexities involved in making the right information available at the right time to power GenAI. Salesforce’s recent announcement of the Unified Knowledge product, available in partnership with Zoomin, reveals their strategy for helping their customers streamline the utilization of corporate data across their platforms.

The Challenges of Data Aggregation and Preparation

Most enterprises have their data spread across many different systems. Integrating information from various external sources into a centralized platform like Salesforce’s Data Cloud has historically been a complex challenge. Data often has differing data formats and resides in siloed systems, making it difficult to create a unified taxonomy and access comprehensive data efficiently.

Additionally, the absence of automated systems for effectively extracting insights from unstructured data previously led to minimal focus on preparing this data for use. However, with the rise of GenAI and its ability to swiftly interpret requests and retrieve relevant information, there is a growing demand for advanced solutions that facilitate easy access to these extensive data repositories. To alleviate the burden of data preparation and aggregation on companies, Salesforce’s partnership with Zoomin is a strategic move, offering sophisticated tools to streamline these processes.

Zoomin’s Role in Enhancing Salesforce Capabilities

Zoomin’s technology facilitates the integration of multiple 3rd party data sources, such as Google Drive documents, AWS S3 assets, Zendesk tickets, websites, other Salesforce orgs, and many others. Beyond enabling these integrations, the Zoomin platform can simplify the complex process of data preparation and integration.

Creating a well-defined taxonomy is helpful for effectively managing and utilizing unstructured data sources within an enterprise. Zoomin’s approach includes several best practices for transforming chaotic data into a structured, accessible format that generative AI can leverage effectively.

Standardization through Taxonomy: By categorizing data into a hierarchical structure of tags, Zoomin enables organizations to standardize the classification of diverse content types. This standardization is crucial for generative AI to understand and retrieve relevant information swiftly.

Enhanced Search and Filtering: The taxonomy serves as a roadmap for both AI and human users to navigate through large volumes of data. Tags and facets defined in the taxonomy allow for refined searches, making it easier to find specific content based on product versions, content types, user roles, and more.

Automated Categorization and Syncing: With Zoomin’s auto-categorization features, documents are automatically classified according to the defined taxonomy. This automation saves significant time and effort, reducing the need for manual data mapping and ensuring that data remains up-to-date within Salesforce’s ecosystem.

To get more specific, Zoomin’s technology offers several features to reduce the manual effort of data preparation. Rather than requiring manual labor to map hundreds of documents into Salesforce’s knowledge base, Zoomin automates and enhances the process through several innovative features:

1. Content Tagging: Users can define specific tags within Zoomin, such as “Product Information,” “Troubleshooting,” or “FAQs.” These tags help in aligning the external data with the structured format of Salesforce Knowledge.
2. Auto-Categorization: Zoomin’s AI-driven tools analyze the content, titles, and metadata of documents stored in platforms like Google Drive or AWS S3 to automatically categorize them according to the predefined tags. This step ensures that data is organized and labeled without manual intervention.
3. Syncing to Salesforce Knowledge: Post-categorization, Zoomin seamlessly syncs the sorted and tagged articles to Salesforce Knowledge. This synchronization not only updates existing entries but also creates new ones where necessary, using the tags to place each piece of content in the correct category or folder within Salesforce.

Sample Case Study: Consider a scenario where a technical manual for “Product X” stored in Google Drive needs to be made accessible via Salesforce. Zoomin automatically identifies this document as a “Troubleshooting” resource (based on a pre-established taxonomy), tags it accordingly, and syncs it with the relevant section in Salesforce Knowledge. This process ensures that when service agents or customers seek information on troubleshooting “Product X,” the AI-powered systems within Salesforce can quickly retrieve and provide accurate and useful information.

Unlocking the Power of Unified Knowledge

The partnership between Salesforce and Zoomin illustrates how CX solution providers are striving to help their customers unlock the potential of their widely distributed knowledge resources. Unified Knowledge is currently in open Beta and slated to be generally available in June of 2024. By automating the intricate tasks of data categorization and integration, Zoomin not only enhances the effectiveness of Salesforce’s GenAI capabilities but also significantly reduces the time and effort involved in data management. However, there remains a need for knowledgeable employees to perform initial tagging and ensure accuracy. When done well, this approach delivers precise information, effectively eliminating “topic creep” and hallucinations, making GenAI-driven service platforms more intelligent and responsive.

Categories: Articles