Akamai: Transforming Data Discovery With LLMs

In 2022, Akamai Technologies, a global leader in cybersecurity, cloud computing, and content delivery services, embarked on a journey to revolutionize its data stewardship and discovery processes by adopting Dataiku, the Universal AI Platform. The newly-streamlined data discovery process built with Dataiku enabled Akamai to save $150,000 in terms of time and resources.

Implementing our LLM-powered chatbot as a one-stop data discovery solution addresses core challenges such as fragmented information, lack of guidance, mistrust in data quality, and time-consuming processes. By enhancing data literacy, improving efficiency, and supporting informed decision-making, our solution empowers organizations to fully leverage their data assets and drive business success. Nirali Dedaniya Data Engineer at Akamai Technologies

The Challenge: Siloed and Outdated Data

The IT Data Intelligence group at Akamai, tasked with simplifying data processes, enabling access, and delivering high-quality, secure, and governed data to fuel insights and operational efficiency, faced several challenges.  Keeping table documentation up to date was a challenge and many tables remained undocumented. Updating documentation was a tedious and time-consuming task that often got pushed towards the bottom of the to-do list.

Additionally, employees faced significant challenges in efficiently accessing and utilizing enterprise data due to fragmented information and lack of clear guidance across various data systems owned by different teams. As a result, business users, application owners, and audit and compliance teams struggled to find the right data and understand its context.

Managing fragmented data systems and outdated documentation made data discovery inefficient, leading to mistrust in data quality and missed opportunities for data-driven decision making.

Discovering Dataiku

Akamai’s journey with Dataiku began with a goal to create a self-service data platform. After evaluating various AI platforms and conducting a thorough request-for-proposal process, Akamai chose Dataiku.

The platform stood out because of its flexibility in supporting both on-premise and cloud installations and its ability to handle the entire AI and machine learning lifecycle — from data preparation to deployment. Dataiku’s balance of full-code and low-code features allowed both technical and non-technical team members to collaborate effectively. Overall, Dataiku provided a comprehensive centralized solution that met the company’s data science requirements.

Akamai implemented Dataiku, internally referred to as the AI360 platform, across several critical use cases. These included improving resource allocation, enhancing security by detecting fraudulent customers more efficiently, better prioritizing support tickets, and more. In 2024, 90% of Dataiku users are data scientists, data engineers, and developers. The remaining 10% are data analysts and business system analysts. Ninety-five percent of users are comfortable with coding and analysis.

A Two-Fold Journey to Efficiency

Akamai needed a solution to streamline the search process and automate the data catalog documentation. There was an urgent need for a user friendly solution that could bridge these gaps and empower users to leverage data effectively. The solution to this challenge was two-fold:

1. Data Stewardship Agent: Akamai’s data stewards historically managed data documentation manually, which often led to incomplete or outdated records. Using Dataiku, Akamai developed a Data Stewardship Agent powered by LLMs that automated the process of generating data table descriptions. By sampling existing data and leveraging available documentation, the agent created detailed descriptions of data objects and columns. This automation significantly reduced the manual workload for data stewards, enabling them to focus on more impactful tasks.
2. Data Discovery Chatbot: Akamai also developed a chatbot designed to streamline data discovery across the organization. This LLM-powered chatbot allowed employees to query enterprise data in natural language, receiving answers and links to relevant data tables, reports, and documentation. The chatbot transformed data literacy within the company by making it easier for users to access and understand the data they needed for decision-making. Whether a business user sought a report, or an analyst needed to understand a metric, the chatbot provided precise answers quickly, enhancing efficiency and building trust in the data.

Benefits of Dataiku at Akamai

By adopting Dataiku, Akamai achieved several significant benefits:

• Enhanced Data Literacy: The chatbot provided a one-stop-shop for all data-related inquiries, improving understanding and utilization of enterprise data across the company. It saved over 6,000 hours company-wide by streamlining the data discovery process.
• Increased Trust in Data: Automated data stewardship and up-to-date documentation built confidence in the quality and accuracy of the data. It saved over 1,500 hours by auto-generating the descriptions of each of the 6,000 datasets, which represents $37,500 in savings.
• Streamlined Data Discovery: The chatbot and Data Stewardship Agent saved time and resources by automating previously manual processes, improving overall efficiency.

There is greater trust in data. Reliable data builds confidence in data quality and usage. By automating data stewardship, we have reduced the manual workload for data stewards by automating dataset descriptions, freeing up our data stewards to focus on more impactful tasks. Santhoshkumar Loganathan Principal Lead at Akamai Technologies

The combination of automated data stewardship and intuitive data discovery tools empowered Akamai to fully leverage its data assets, driving business success and improving decision-making across the organization. With these solutions, Akamai continues to enhance its operational efficiency and maintain its position as a global leader in technology.