The Qlik Agentic experience introduces a unified conversational interface for working with data and analytics, delivering fast, explainable, context-aware responses through Qlik Answers. It also enables customers to securely use external LLMs with trusted Qlik data with our MCP server, and includes specialized agents that automate analysis and action—while allowing users to stay in their preferred AI interface.
And it is, of course, underpinned by Qlik's core strengths:
Trusted data products built from high-quality, governed data
An open agentic architecture that allows both Qlik agents and third-party AI assistants to tap into the full range of Qlik capabilities
With this launch:
Qlik Answers becomes your unified agentic AI assistant, reasoning across structured and unstructured data with the power of AI’s knowledge and capability to deliver insight, assistance, and automation.
Qlik MCP server delivers the power of Qlik to third-party AI assistants for building custom agentic solutions powered by governed data and our unique analytics engine.
Discovery Agent will provide always-on anomaly and change detection, proactively surfacing meaningful, high-priority outliers and anomalies through a simple insights feed. Available in March.
Data products in Analytics extend data products and quality capabilities into the analytics workflow, strengthening the foundation for agentic AI. Available February 24th.
You can read about all four launch highlights in our Innovation Blog post, Qlik’s Agentic Experience is Here, where we'll walk you through the details on each.
And when you're ready to get started and learn more, explore our resources:
I've been playing around with Answers and there's definitely a lot of stuff going right, but also a decent amount of stuff going wrong. However, nothing is as egregious as the response times. Hopefully that's something that's being prioritized on the R&D/development end.
Incidentally, is the Answers team looking for feedback? If so, how does one provide it?
Yes, response times is actively being worked upon and we are actively looking for feedback.
If you would like to share your thoughts, please feel free to reach out to me directly via DM or send me an email to adithya.pai@qlik.com with your feedback. I am very happy to review it and discuss further.
Last year, we built a GPT in ChatGPT that was fed with data uploaded from Excel (across multiple areas). Thanks to that, some of our directors were able to perform much deeper analyses on the information.
By default, we use it with a thinking model, and in my experience the ~90-second response times they mention are absolutely worth it if the quality of the analysis improves substantially—as we’ve already confirmed with thinking models. My users are thrilled with the results.
More recently, I’ve been testing MCP through ChatGPT, and something very similar happens: with a non-thinking model the response times are almost instantaneous, but the analysis becomes significantly deeper when switching to a thinking model. Personally, I’m completely comfortable with those response times.
I’ll continue testing, but I think we could migrate our GPT from manually uploaded data to one connected via MCP. I haven’t had the chance to try Answer yet because it’s not enabled in my Qlik Sense SaaS Enterprise version; according to the FAQ, it should be available starting February 24.
