AI Assistant for Enterprise

My Experience with Amazon Q Connectors

Category: genai
Category: genai

AI Assistant for Enterprise: My Experience with Amazon Q Connectors

In today’s fast-paced business environment, having quick and easy access to information is crucial. Enterprises often rely on multiple data sources, such as public websites, Google Drive, Slack, and GitHub, to store and manage their documents. Importing documents from these various sources into a single, centralized AI assistant like Amazon Q can significantly streamline operations. By doing so, employees can effortlessly retrieve information, reducing the time spent searching through different platforms. This integration can effectively replace traditional intranets, offering a more efficient and interactive way to access company data.

An AI assistant, also known as a chatbot, is a software application that uses artificial intelligence (AI) to simulate human-like conversations with users. These assistants are becoming indispensable in enterprise settings, streamlining processes and enhancing user experiences. Today, I’m sharing my journey with Amazon Q, an AI assistant announced during re:Invent 2023, focusing on its connectors to create an enterprise-specific chatbot.


Unveiling Amazon Q


Amazon Q was in its preview mode when I started exploring it. While some of the challenges I faced may no longer exist, they provide valuable insights into the setup process. My goal was to feed current company data into Q and create an internal chatbot, focusing on the following connectors:


  • Public website
  • Google Drive
  • Slack
  • GitHub


Step 1: Creating Amazon Q and Setting Up Authentication


The first step was to create Amazon Q and define the authentication mechanism. For ease of use, I opted for Google Login for employees.



Step 2: Connecting Data Sources


Each data source required a unique type of authentication, and I had to figure out how to create the necessary tokens as the documentation for Amazon Q didn’t provide end-to-end instructions.


Credentials for each data source were securely stored in AWS Secrets Manager.



Step 3: Integrating GitHub


Integrating GitHub was straightforward with the right token and credentials stored in AWS Secrets Manager.


Step 4: Tackling Google Drive


Google Drive presented a challenge. I had to create an application under the Google Admin console and define a read-only role. The credentials were again stored in AWS Secrets Manager.



The store the credentials again under AWS Secrets Manager:


Step 5: Adding Slack


Adding Slack completed the setup, though it required careful handling of tokens and permissions.


Challenges and Resolutions


Even though the sync status occasionally showed failures, I found that documents were being indexed. However, the error logs were not detailed enough to pinpoint the issues. Here are some key takeaways from support responses:


  • Interval Management: Keep the sync interval narrow. Google API and Slack API may block high-volume document crawls.
  • Connector Limitations: These connectors are not yet optimized for high-volume crawling, which needs addressing in future updates.


The Chatbot Experience


Once the documents were indexed, the chatbot experience began to take shape. Here’s a glimpse of the final product:




For my use case, Amazon Q, with its robust set of connectors, shows great promise in replacing internal intranet/company portals for employees. However, it’s crucial to consider the following limitations upfront:


  • Language Support: Amazon Q Business is optimized for English.
  • Connector Documentation: The documentation does not cover all scenarios in detail, particularly for high-volume environments.


Despite these challenges, Amazon Q is a promising tool for enterprise AI assistants, and with future updates, it has the potential to become even more powerful and user-friendly.


Stay tuned for more updates and insights on using AI assistants in the enterprise!

Derya (Dorian) Sezen

Derya, a.k.a. Dorian, ex-CTO of an subsidiary, is currently working as Cloud and DevOps Consultant at kloia.