Qlik Says Its Discovery Agent Has Surfaced More Than 100,000 Discoveries

Qlik’s enterprise AI tools are drawing continued use, according to an AWS case study published October 7, 2026. Most Qlik Cloud accounts with agentic tools enabled actively use them, it says. The disclosure also explains how Qlik Answers routes workplace questions and checks generated answers against source material, rather than relying on document retrieval alone.

The case study was co-written by two AWS solutions architects and Qlik engineering director Kent McLean. It describes adoption since Qlik Answers became generally available in February 2026—not a new launch. Qlik’s Discovery Agent, which detects anomalies and outliers, has surfaced more than 100,000 discoveries since launch, according to the authors.

A router before the reasoning

Qlik Answers gives employees one conversational entry point, but the work behind it is divided. A lightweight routing layer reads the question and conversation context, then chooses where to send the request. It does not try to solve the task itself.

An answer layer then chooses a fast path for simple requests or breaks harder questions into smaller ones. Structured data questions go to a dedicated analytics path. Document questions use Amazon OpenSearch Service to find relevant content. Specialist agents share a common framework for tools, working context and human-review steps.

Retrieving a document is not the last check. Qlik compares generated answers against the source content using Amazon Bedrock Guardrails. The model-access layer also filters requests and responses for prompt injection, personal information, secrets and denied topics. Retrieved document context is carried into the final answer so it includes citations.

75% fasterLintech response times

The case study says Lintech International indexed more than 17,000 technical documents and returned up to seven hours a week to business managers.

Scaling without one global deployment

Regional availability adds another constraint. Qlik says Bedrock’s cross-Region inference supports deployment across 11 Regions while meeting data-residency requirements. When a needed model is unavailable locally through Bedrock, Qlik hosts it on Amazon SageMaker AI, then moves the workload back when Bedrock availability catches up.

Qlik also forecasts model capacity three to six months before major launches, projecting token consumption by feature and Region. It checks those projections against actual usage after rollout. New capabilities can be enabled selectively for individual customer accounts, rather than changing the default experience for everyone at once.

Other customer examples show different uses: Bystronic deployed a chatbot in 15 minutes, while TouchPoint Support Services uses Qlik Answers to provide compliance-aligned guidance to 15,000 staff across 650 healthcare sites, according to the case study. These are vendor-reported outcomes. The adoption disclosure gives neither the exact account share nor a definition of active use.