Are you tired of dealing with repeated questions that clog up your Slack channels? Good news! After working with early users for the last few months - today - we are thrilled to unveil the Beta version of our Automated Answering System. This state-of-the-art AI chatbot technology is designed to seamlessly integrate with your Slack-based helpdesk and customer support systems, revolutionizing the support experience for customers and employees alike.

‍

The Challenge: Slack Channels Are Overflowing

Customers and employees love Slack. It's much easier and faster to collaborate, without the need to switch to a different tool. Many of our customers have hundreds of channels with customers and busy internal channels fielding thousands of questions a month. ClearFeed's software helps track and close all these questions, alerts on SLA misses, integrates Slack with systems like Zendesk, Jira & Freshdesk, and provides detailed analytics on Slack activity.

However, the ease of using Slack can also lead to repeated questions that take up valuable bandwidth of service teams. Support teams, of course, have known about this problem for years. Aside from documentation, they have also invested in preparing FAQs and KBs for their customers. However, users prefer asking questions in Slack instead of looking up support portals for answers.

In today’s fast-moving world, service teams have also struggled to update their knowledge bases. Products and Interfaces are constantly evolving. Ironically, Slack channels themselves are one of the best sources of knowledge - but are both hard to search easily and contain confidential data that cannot be shared freely with other customers and employees. 

Most organizations also have a diversity of data sources - some public and some private - that can be leveraged to help customers and agents. Customer portals, even when used, can help users with public documentation and KBs - but by themselves do not help agents who may have access to private troubleshooting guides and documentation that is not public.

The advent of chatGPT and related AI systems has also been a turning point for user expectations. Everyone is asking:

‍ This is where ClearFeed's automated answers come into play.

‍

The Solution: ClearFeed's Automated Answering System

Our Automated Answering System is more than just an AI chatbot, it’s a system that: 

• Indexes your public and private knowledge sources: Including chat on Slack.
• Provides an agent assistant: Admins can configure "Suggested Answer" workflows for agents using both public and private knowledge sources.
• Deflects common questions: Suggest answers and documentation links to customers directly and enable problem resolution without agent intervention.
• Provides strong privacy controls: Different knowledge sources can be configured for for different groups of customers and agents.
• Provides KB search within Slack channels
• Supports federated search: As a bonus, ClearFeed can also use data in existing Search backends like Algolia, Elastic Search, and more. Enterprises can leverage their existing investments in enterprise search and plug it into their Slack-based helpdesk and customer support channels.

‍

Making a Difference in Production

Deployed at many ClearFeed customers - our automated answering systems are already speeding up the quality and speed of agent responses on common customer questions. As Shravan Ashok, Principal Program Manager at CastAI, put it:

‍

How It Works

ClearFeed indexes organizes, and leverages different knowledge sources in a company and leverages them using generative AI in a Support context.

‍

‍

Vector Indexing for Quality Search Results

ClearFeed indexes Support related content like Documentation, Wiki pages, FAQs, and Knowledge Base articles in modern vector databases. Admins can authorize and configure ClearFeed to periodically crawl and index these data sources. ClearFeed can handle a wide variety of data sources and ensures that their data is indexed to provide high-quality search results.

‍

‍

GPT-Powered Automated Answers

ClearFeed can be configured to provide automated answers in various contexts. When so configured, it retrieves relevant search results from the index and uses OpenAI's GPT-4 and other state-of-the-art AI models to suggest answers automatically. 

As the answers are generated by using the configured knowledge sources - the answers are grounded in truth. By providing truthful answers only, it ensures that users have a high level of trust in the system. Every answer carries a citation to the original source - further reinforcing trust and empowering users with relevant original sources at the same time.

‍

‍

Context-Specific Configuration

ClearFeed makes it easy to configure different knowledge sources while generating answers for customers and agents. For example:

• Agents can use both private and public data sources.
• Customers may only be able to use public knowledge sources.

Furthermore - depending on the customer - only documentation for a specific product or version may be relevant. Administrators can set up different knowledge sources, tag them, and use tags to specify sources that can be used in specific contexts.

This level of customization ensures that every interaction is tailored to the individual, providing a more efficient and targeted support experience.

‍

Inline Search Experience

Where automated answers are not possible, or available, users also have the option of looking up Search results. Agents can edit and paraphrase customer questions and get desired answers and search results without leaving their Slack workspace. A sample flow is shown below where the user can browse through all the matching search results - and also edit the request and perform search against the modified request:

‍

‍

Capturing Knowledge in Slack

ClearFeed tracks requests and responses in your Slack channel. Using the power of generative AI - it is able to synthesize new knowledge base articles from Slack threads that are not covered in existing documentation. Support teams can review and edit proposed articles before adding them to their knowledge bases - and also keep track of all pending review items.

Capturing knowledge continually and using that to improve the quality of search continuously provides a powerful positive feedback loop that makes the automated answering system better over time, improving both the quality and coverage of the answers.

‍

Conclusion

At ClearFeed, we have been early adopters of AI, with features like automated request classification, categorization, and conversation boundary detection all powered by deep learning models. Building an AI-generated answers system aligns with our vision of bringing NLP technologies to everyday technology workers and promises a transformative experience for support teams and users alike. By integrating modern databases and foundational AI models like GPT-4, we're pioneering a new era in support where responses are not only swift but also tailored to individual needs.

Ready to revolutionize your support experience with automated answers? Contact us at support@clearfeed.ai, join our community Slack channel, or book a demo with us.