Implementing an AI chatbot to enhance data team efficiency in a remote organization from Coalesce 2023

"We're able to hook up this ChatGPT-based chatbot directly to our business tables and leverage all the existing work that we've already done."

- Alex Gronemeyer, Lead Analytics Engineer at Airbyte

Team members from Airbyte, Alex Gronemeyer, Lead Analytics Engineer, and Anil Maharjan, Senior Data Analyst, discuss their implementation of an AI chatbot named Dot. Airbyte developed it to streamline data analysis and answer data-related queries. They also discuss their experiences from the initial stages of vendor analysis to the final rollout of the AI chatbot.

AI chatbots can effectively support data teams

The introduction of an AI chatbot named Dot to the data team at Airbyte significantly improved their operations. Dot, the third member of the team, was designed to handle a bulk of inquiries, saving time and increasing productivity.

Anil explains the integration and functionality of Dot, stating that "the third member of our team is what we like to call our Junior Data Analyst, Dot. Not only is Dot cute, but Dot is an AI chatbot that is supporting the growth of our data team."

According to Alex, Dot helped manage a high volume of data questions, addressing the team's problem of having "too many data questions to answer and not enough people or time."

Balancing the need for AI and privacy

Introducing an AI into their data operations raised concerns about privacy and data security. So, the team had to find a way to use Dot without compromising sensitive data.

Alex states, "We take security very seriously at Airbyte… one of the more time-consuming parts of this process is figuring out which vendors we felt were secure enough to even start integrating our data with."

To tackle this, Alex explains that they created a system to anonymize data at the staging level in dbt and created dbt tests to identify any sensitive data that needed to be masked.

AI could be the solution to scaling data operations

Despite the challenges they faced in integrating AI into their operations, the Airbyte team remains optimistic about the potential of AI to scale their data operations.

Anil states, "Can we scale our data with an AI chatbot? We think so… we've saved a lot of time on writing ad hoc queries… It's enabled people outside of our data team to interact with the data, and finally, it's given us visibility into the questions being asked."

While there are still issues, such as the stability of new technology and the discovery of raw data, the team sees AI as a viable solution for managing the volume and complexity of their data operations.

Anil and Alex's insights

• AI chatbots can effectively handle a high volume of data-related questions, saving time on ad hoc queries
• It's crucial to consider factors like the nature of the questions, the data sources, and how the team will interact with the bot when selecting a vendor
• Security and data anonymization are significant aspects of the process
• The AI chatbot can function effectively in a platform like Slack, providing answers within seconds and facilitating data interaction
• Despite some challenges, such as occasional outages and incomplete documentation, the AI chatbot has been a success, enabling people outside the data team to interact with the data