Qlik announced on 17 November 2026 that Qlik Answers, its AI assistant for analytics, now supports custom skills and a set of advanced analytics: prediction, optimisation, causal analysis and what Qlik calls gray, or excluded-set, analysis. The same release moved Qlik Answers onto a single agent that can build charts and sheets from plain-language requests, and added a score that tells app owners how ready their app is for AI questions.
What changed
Custom skills. A skill is a short set of instructions an app owner attaches to an app or an assistant. It can hold your own terms, the way a metric should be calculated, or the steps an analyst would follow to answer a common question. If "active customer" means a purchase in the last 90 days in your business, you write that down once and Qlik Answers uses it every time. Qlik says setting one up takes under an hour.
Advanced analytics. Users can now ask forward-looking and comparative questions, not only "what happened". Prediction estimates likely outcomes. Optimisation and simulation weigh options against constraints. Causal analysis looks at what is driving a change. Gray analysis is the one Qlik users will recognise: it looks at the data your current selection leaves out, which is where the surprising answer often sits. All of this runs on the Qlik Analytics Engine, with the language model handling the conversation.
One agent, more actions. Qlik Answers can now create and edit charts and sheets from a description and handle requests that take several steps in one conversation. Under the hood the service runs on Amazon Bedrock, and Qlik says existing deployments keep working while the change rolls out.
An AI readiness score. Each app gets a score showing how well prepared it is for Qlik Answers. In practice that means clear field names, descriptions and master items, the same housekeeping that makes an app easier for people to use.
Qlik describes the new features as available now in Qlik Cloud, with the usual note that some functionality rolls out in phases and may not reach every region at once. On the same day it also expanded the Qlik Predict tools exposed through its MCP server.
Why it matters
The weak point of most AI assistants in analytics is meaning, not maths. A model that does not know your fiscal calendar or your definition of churn will give a confident wrong answer. Skills move that knowledge out of individual analysts' heads and into the app, where it can be reviewed and versioned like any other piece of business logic.
The advanced analytics matter because they shift the assistant from reporting to decision support. A supply chain lead who can ask "what happens to fill rate if we cut safety stock by 10% at these two depots" is doing work that used to need a data scientist and a week. That is useful, and it is also where errors become expensive. Predictions need someone accountable for checking them.
The readiness score is a quiet but helpful addition. It turns "is our data ready for AI?" from a workshop topic into a number an app owner can improve.
What this means for organisations in the Middle East, India and Africa
Plenty of Qlik estates in the Gulf, India and South Africa still run Qlik Sense client-managed on their own servers. These features are in Qlik Cloud. If you are still on premises, the question is no longer whether to move but which apps to move first.
Running on Amazon Bedrock raises a data residency question. Check which Bedrock region your Qlik Cloud tenant uses, and whether prompts and app data leave your chosen region, before you switch Qlik Answers on for apps holding personal data. Under the UAE PDPL, Saudi PDPL and India's DPDP Act you need to be able to say where personal data is processed. Bilingual organisations should also test how well skills and answers handle Arabic or Hindi field names and questions before rolling out widely.
What to do next
- Pick two high-traffic Qlik Cloud apps and check their AI readiness scores. Fix field names, descriptions and master items first.
- Write three or four skills for each app, covering the metrics people argue about most. Have the metric owner sign them off.
- Trial gray and causal questions with one business team, and compare the answers with what an analyst would produce.
- Before using predictions in a decision, agree who checks them and how often.
- Confirm tenant region, Bedrock processing location and logging with Qlik in writing, then update your records of processing.
For help planning a move to Qlik Cloud or setting up Qlik Answers, see our Qlik partner page and our data analytics services.