In association with Qlik
As businesses face manpower crunches and pressures to deliver faster, more personalised service, agentic AI has become a goal that many seek to give them a competitive edge for the future.
Yet, most do not have the data foundation to enable the analysis, automation and reasoning that the next leap forward in AI promises.
Despite the enormous hype of AI, most businesses do not have the technological maturity to reap its true benefits. While an overwhelming majority – 80 per cent – of businesses in Asia-Pacific expect to invest in agentic AI, only 12 per cent are confident in their data, according to research firm IDC.
Key to overcoming this issue is finding and making sure data is of high quality and can be trusted, say experts. With this foundation, they note, businesses can truly take the big leap into the future with agentic AI.
A giant step forward
However, that next step has to be grounded in good data foundation. AI agents acting semi-autonomously need to have the right information to reason and make a decision or recommendation.
Several AI agents can work together to help a traveller book and adjust his ground transport according to his flight arrival times but what if the arrival times for the flight differs between different sources? Which source should the AI trust?
Trust in quality data is the most important part of agentic AI, said Maurizio Garavello, the senior vice-president for Asia-Pacific for Qlik, which provides data analytics and integration, as well as AI tools to businesses.
It is important to know where AI is basing its decision on and where the data is coming from – a trusted sources or a YouTube video, he pointed out, during a recent analyst and media event in Singapore.
“Data is the new oil but it may be in the sand, it has to be taken out and refined and be made ready for use,” he noted.
Trusted framework
As a company focused on data, Qlik offers a way to build a trusted foundation by curating not just structured data but also trawling the unstructured data available to a business.
About 80 per cent of today’s data is unstructured – in the form of Excel files, images or videos – and a good data foundation needs to not only ingest the information but try to reconcile the differences to have a single source of truth.
For example, Qlik blends both structured and unstructured data with its indexing engine and helps generate answers that overcome common issues that many large language models (LLMs) face today, such as with numbers.
Instead of mixing these numbers up, Qlik enables responses that are accurate when presented to a user, for example, a marketing manager seeking suggestions to boost sales in a product.
The rich responses that Qlik delivers includes videos and graphics that are interactive, so users can not only view a nice dashboard but also take action directly by interacting with the information, said Charlie Farah, Qlik’s field chief technology officer for analytics and AI.
A key difference is that users can quickly check the source of the data and find where the AI has based its response on, he added. “Explainable AI is key to trust.”
“You can’t do AI without real data – trusted data,” said Garavello. Humans matter too, he added, especially when it comes down to breaking down silos.
Sales and finance, for example, may have different numbers for the revenue from a product and there is a growing need for data stewards that ensure that data is of high quality before it is fed to an AI engine.
Just as important is ensuring there are guardrails in place. Kelly Forbes, who is on Qlik’s AI Council, said many businesses are thinking of how to deploy AI safely and implement business cases in the local context.
It is important to build the foundation rather than be led by competitors, she noted, during a panel discussion at the Qlik event.
Also speaking at the event, Professor Quek Khor-Ping, deputy director of the business analytics centre at the National University of Singapore, said a governance system would have to move in tandem with the fast pace the technology is evolving.
Key to this is the question of when or where a “human in the loop” has to be involved in making a decision, he noted.
Now is the future
Despite the challenges, it is clear to businesses in Asia-Pacific that AI has to be the way forward. While challenges exist today for many early adopters, what they have from their experience is an idea of what works to build a valuable foundation to accelerate into the future.
Sometimes, it takes the right tools to unlock the potential. What differentiates Qlik from other data or AI players is its underlying Qlik Associative Engine that enables AI to think more like humans do.
That is, understanding what a term or subject means – “apple” should fire up associations with a crunchy fruit, Steve Jobs’ company or a Beatles’ recording. These associations are picked up by AI that runs on Qlik’s engine, which also tries to make sense of the content through context.
Instead of forcing users to adapt to computer logic, this approach mirrors human cognitive processes and delivers better results.
Garavello pointed to a client that had not managed to make full use of its data previously until Qlik’s solution helped it create a trusted source that AI can use to predict customer patterns and the need for temporary staffing.
The theme park and cinema operator, he said, saw improvements immediately after just two days using Qlik Answers, which gave it accurate predictions of customer arrivals and enabled it to optimise staffing cost while delivering an improved experience.
“We always ask customers to challenge the data and be curious and ask the AI,” he noted.
Find out more about Qlik’s solutions to enable the next big leap into agentic AI here.
