Qlik AutoML update targets trust with visibility, simplicity
The vendor added new features that simplify developing and deploying machine learning models as well as provide greater visibility into performance to ensure trustworthy outputs.
Qlik on Tuesday updated AutoML, adding new capabilities designed to foster trust in machine learning models by simplifying development and deployment and providing greater visibility into their performance.
Trust has been an ongoing theme for Qlik over the past few months as it creates an environment for its customers to develop and deploy AI models and applications.
In June, the vendor unveiled Qlik Talend Cloud, a data integration platform aimed at ensuring the data on which AI tools are trained can be trusted. Simultaneously, it introduced Qlik Answers, a generative AI-powered assistant designed to provide users with trusted responses to natural language queries so that they can confidently take action.
Both were made generally available in July.
Now, trust is a focus of the latest Qlik AutoML update. First launched in August 2022 after Qlik's 2021 acquisition of Big Squid, AutoML is Qlik's suite for developing, deploying and serving machine learning models and is part of Qlik's Staige portfolio for developing AI tools.
With enterprise interest in AI and machine learning on the rise, and organizations entrusting AI and machine learning models to take on certain tasks previously performed by human beings, being able to trust AI and machine learning models is critical. Qlik's addition of features to AutoML that engender trust is therefore significant, according to Mike Leone, an analyst at TechTarget's Enterprise Strategy Group.
Trust is so important because the effectiveness and ultimate success of these technologies depends on their ability to give reliable and understandable results.
Mike LeoneAnalyst, Enterprise Strategy Group
"Trust is so important because the effectiveness and ultimate success of these technologies depends on their ability to give reliable and understandable results," he said. "People need to feel confident that the models are making accurate predictions or giving accurate responses backed by solid reasoning."
In addition, trust leads to higher user engagement, with workers more likely to continue using tools they believe in, as well as more workers beginning to use the tools if they know they're trustworthy, Leone continued.
"End of the day, establishing a foundation of trust ultimately leads to broader adoption and better business outcomes," he said.
Based in King of Prussia, Pa., Qlik is a longtime analytics vendor that in recent years complemented its BI platform with data integration capabilities and is now focused on adding an environment for AI and machine learning. Toward its goal of providing customers with the capabilities to create AI applications, Qlik in January acquired Kyndi, a specialist in unstructured data management.
New AutoML capabilities
Qlik's AutoML update comprises four new features.
Intelligent model optimization is designed to simplify model development and deployment. The feature automates parts of the development process to remove some of the burden from developers, data scientists and others involved in the development process, applying pretrained best practices that result in better model performance while reducing the effort it takes to build models.
Native machine learning analytics enables developers and other machine learning model stakeholders to observe the performance of models. The feature provides users with an autogenerated dashboard that displays details about a given model so that users can get insights into what the models predict and what drives those predictions.
A comprehensive set of machine learning operations capabilities aims to ensure the model accuracy that leads to trust in model outputs. Among the MLOps capabilities are automated monitoring for issues such as model drift, model retraining and lifecycle management.
Finally, full integration between Qlik AutoML and Qlik Cloud aims to give users easy access to their data infrastructures to provide them with a simple user experience that results in efficient data-informed decisions.
At their core, the new capabilities aim to provide enterprises with more trust in their machine learning outputs, according to Brendan Grady, general manager of Qlik's analytics business unit.
"The latest Qlik AutoML enhancements provide greater visibility, control and easy model management, which directly foster trust," he said.
For example, automated monitoring for model drift enables users to quickly understand when models start losing their accuracy so that users can retrain the models to restore their reliability. Similarly, autogenerated dashboards that provide users with details about model performance let those users know if models are performing at a trustworthy level or need to be adjusted.
"These capabilities lay the foundation for future automation in model lifecycle management, further enhancing reliability and confidence in AI-driven decisions," Grady said.
Leone similarly said the new AutoML features have the potential to accomplish Qlik's goal of providing users with greater trust in machine learning model outputs. In particular, the features that provide explanations about model performance engender belief on the part of users that they can take actions based on what models are telling them.
"The focus on model explainability means users can grasp the reasons behind predictions, building trust in the insights they get," Leone said. "This update serves as a mix of accessibility, transparency and advanced analytics to not only accelerate data-driven strategies, but also helps Qlik stand out as a continued leader in the analytics space."
In terms of the individual features, the MLOps capabilities and model optimization tools stand out, Leone continued. Individually, they provide users with insights. Beyond that, they complement one another by creating a framework for developing, deploying and maintaining machine learning models.
"These capabilities can help organizations respond swiftly to shifts in data, but more it reduces the time to market for new models while maintaining reliability and transparency," Leone said. "This is what organizations are asking for, so it's a great fit for where most of the market is today."
While trust is an underlying tenet of Qlik's AutoML update, customer feedback played a role in determining which features to develop, according to Grady.
Customers use Qlik AutoML for predictive analysis, he noted. As they do so, they have asked for better ways to understand their model outputs. However, part of the impetus for the AutoML update came from Qlik's own observations about the future needs of business, Grady continued.
"Development is ... future-looking, anticipating what businesses will need to stay competitive," he said. "For example, we foresee growing demand for proactive decision-making and increased automation as AI adoption accelerates."
Coming up
While trust has been a guiding principle for Qlik in recent months, a need going forward is enabling customers to use not only structured data to inform analysis, but also unstructured data.
Analytics has historically focused on structured data such as financial records and point-of-sale transactions. Now, however, unstructured data such as text, images and audio files makes up the vast majority of all data -- perhaps over 80% of all data.
Tapping into that unstructured data, therefore, would provide enterprises with a more complete view of their business.
But beyond traditional analytics, AI -- and generative AI, in particular -- requires large amounts of data to be accurate and minimize incorrect and sometimes even offensive outputs called AI hallucinations. An organization's structured data alone might not be enough to train an AI or machine learning model.
In response, many data management vendors have added vector search and other capabilities that enable customers to access unstructured data and combine it with structured data to provide more complete information about an enterprise, as well as provide that enterprise with the proper means for training models and applications.
Qlik has historically added capabilities through acquisitions. For example, its data integration platform was developed in large part by acquiring Attunity, Podium Data and Talend. Similarly, AutoML resulted from Qlik's acquisition of Big Squid.
The vendor's acquisition of Kyndi provides it with a foundation for tapping into unstructured data.
Mike Capone, Qlik's CEO, said in June that enabling customers to access unstructured data and combine it with structured data is a priority. Grady echoed that.
"Over the next six to 12 months, Qlik is focused on enabling businesses to gain a competitive edge by deepening their understanding of both structured and unstructured data," Grady said. "We are working on advancing our capabilities to provide more comprehensive insights, not just for data scientists, but for everyone in the organization."
Eric Avidon is a senior news writer for TechTarget Editorial and a journalist with more than 25 years of experience. He covers analytics and data management.
