Day 05
Bringing the power of your data to AI
Achieving excellence in AI with the right data
Watch the Recorded Session on Bringing the Power of Data to AI
Customer case
How HARMAN makes smart decisions in real time
From exploration to activation
HARMAN, an independent subsidiary of Samsung, is a global leader in connected car technology, lifestyle audio innovations, professional audio and lighting solutions, and digital transformation. While Samsung drives the company’s mass data governance and analytics strategies, HARMAN’s data analytics team manages analytics for the entire company, including ingestion, maintenance and governance.

You may have a HARMAN digital cockpit in your car, a HARMAN Kardon home theater in your living room, or JBL earbuds for your daily commute. Touring musicians use our sound and lighting equipment in concert, and many of your favorite albums and singles were recorded with our pro audio gear. And many enterprises take advantage of our digital transformation solutions to achieve better business outcomes using emerging tech such as generative AI, cloud, and advanced analytics.

In 2019 various departments of HARMAN were using QlikView, but they were developing applications in silos with no single analytics standard to ensure continuity and consistency across departments. Individual teams were building apps that were specific to their operations, and they didn’t scale them across the company. The development of analytics apps in silos meant business users spent a lot of time finding, ingesting, and validating the data going into these tools — users spent 80% of their time ingesting data and 20% analyzing it. To get business benefits, HARMAN data analytics team wanted to flip that ratio.

The migration project to Qlik Sense had already begun when Annette Jonker, Senior Director, Data and Analytics at HARMAN International, joined the team. They formed a centralized data and analytics team to affect the transition and consolidate data analytics usage and development. Qlik Sense was the perfect solution — it was scalable, and the company could leverage its existing QlikView knowledge to get the most out of the platform.
Qlik has given HARMAN visibility into our operations and those of our suppliers. We can see what’s happening now and what’s coming down the line, so we can strategize and plan more effectively.
— Annette Jonker,
Senior Director, Data and Analytics, HARMAN International
Some of HARMAN’s use cases include sales, inventory, and supply chain analyses. The company leverages Qlik across internal operations, from sales to legal and finance, and the results are highly impactful. To achieve the optimal benefits, the team ingests data that is granular enough to provide full visibility into all operations. When users can drill down to the transaction level, it increases trust in the data and allows everyone to generate more meaningful insights.

The team identified ChatGPT as an enabling technology that could be used to further accelerate the process of turning data into insights. With this in mind, HARMAN conducted a pilot project to embed GPT functionality into Qlik dashboards. The first solution uses VizLib’s library of third-party visualisations.
As the data was already available in Qlik, they were able to leverage Generative Chat to address unstructured queries related to that data. The success of this approach led the team to develop a private integration with generative AI, which is now being piloted for broader use.
The team is also working closely with HARMAN’s Digital Transformation Solutions group to integrate generative AI and Qlik, pushing the platform’s boundaries by moving beyond descriptive analytics and empowering people to gather contextual insights from data using natural language questions — even if they’re not data scientists.

Once users have posed their questions, ChatGPT is responsible for providing the answers. The natural language queries and datasets are sent to ChatGPT via the OpenAI platform. The process is transparent to end users because the front end is a Qlik dashboard, which eliminates the need for them to write SQL.
Previously, people would ask, "What is AI/ML?" Now, the question is, "When and how can we use AI/ML?" It's definitely an area of growth, especially now with Qlik’s Auto ML codeless automation, which allows us to run experiments and probe deeper.
— Annette Jonker,
Senior Director, Data and Analytics, HARMAN International
HARMAN’s mission is to enhance quality of life and connect people to what they love, wheverer they are. Analytics-driven decision-making is a key enabler for greater adaptability. From supply chain management, production efficiency, and profitability maximisation, it’s essential to adopt a proactive rather than reactive approach.
Additional theory
Build generative AI on the right data

Support AI Ready data

Artificial intelligence (AI) is a field of computer science that attempts to create systems that can mimic and outperform human intelligence. A subset of AI is machine learning (ML), where computers learn from data and are able to make predictions without programming. One type of machine learning is deep learning, which uses artificial neural networks to process information and learn in a way inspired by the human brain. Finally, within deep learning, we have generative AI, which focuses on generating new content.

Generative AI, or GenAI, is an AI system that can generate text, images or other data using generative models in response to human input. GenAI models learn the patterns and structure of their input training data and then generate new data with similar characteristics.

In Generative AI, data can't come second. Instead, it is the core fuel which drives an organisation's ability to create business value from generative AI.

Organisations that want to use AI must create a data environment that is ready for it. This means more than just collecting lots of data or investing in the latest AI tools. It means creating AI ready Data: making sure that data is managed, governed and used in a way that aligns with the principles of clarity, quality and accessibility.

Data should be unbiased

The term 'AI bias' refers to the potential for biased results in machine learning and algorithms due to human biases in the original training data or the AI algorithm itself. These biases may lead to distorted outputs and potentially harmful outcomes.

To make sure AI isn't biased, organisations must have AI governance. In essence, AI governance is the creation of a set of policies, practices, and frameworks to guide the responsible development and use of AI technologies. AI algorithms can be black box systems that use data in an unfair way. Transparency practices and technologies help ensure that data is used to build AI systems fairly.

Data should be fresh

Real-time data is a continuous flow of data in motion. It is streaming data that is collected, processed, and analysed on a continuous basis. In order to meet the demands of today’s fast-paced business environment, companies must have the necessary technologies, such as Change Data Capture, Stream Data Capture, and Continuous Data Processing.

Data should be accurate

AI-powered systems require the use of high quality data to be fully effective. It is essential to continuously assess the organisation's data assets against key data quality metrics such as relevance, reliability, accuracy, etc. Data quality tools and technologies should be seamlessly integrated into existing workflows.

Data should be secure

The use of masking, tokenisation and access control techniques is essential to protect access to sensitive data and to carefully control the movement of data outside the private network. It also requires active engagement and education at all levels to embed ethical data use and security principles into the organisation's data culture.

Data should be discoverable

For AI readiness, data accessibility is critical. Silos and access restrictions can significantly limit AI systems' ability to generate insights across an enterprise. Establishing clear roles for data ownership as well as using data catalogs and indexing data allows companies to improve data discovery and use.
Qlik has been at the cutting edge of data and analytics advancements for the past 30 years, leading the way with AI in a platform. Today, it delivers a full range of augmented analytics capabilities for deeper insights, as well as automated machine learning to easily build models and generate predictions.

Moreover, the latest acquisition of Kyndi directly addresses the rapidly growing volume of unstructured data in the world. Qlik's incorporation of Kyndi's technology into its cloud solutions enriches decision intelligence by providing more comprehensive answers with enhanced context and relevance. This advancement strengthens the management and curation of enterprise data, fostering trust and consistency across organizations while upholding stringent governance and security standards.

Finally, Qlik Talend now offers a full stack of data integration, quality and governance capabilities to help organisations build the most advanced data fabric for supporting your business. This integrated data platform can act as the primary information, insight and action facilitator for generative AI so that it can be confidently applied across the organisation.
Creating a data preparation in Talend Cloud
Sign up for a Talend Cloud free trial account, if don't have one. Step-by-step guide on accessing Talend Cloud free trial is published in the Day 1 of Data Integration Week.
Related Links
10 trends in data, analytics and AI that organizations need to act on today.
In this e-book, you’ll learn how to build AI- and ML-enabled data pipelines for delivering AI-ready data.
A series of video episodes to discover practical guidelines for data and analytics professionals.
Staying in touch with Qlik Data Integration and Analytics
subscribe to biweekly LinkedIn Newsletter "Data Matters" with news, events and cases from CEE market