In partnership with

🔍 Deep Dive: Building an Assistant with Qlik Answers in Qlik Cloud

Qlik Answers lets you build an enterprise-grade knowledge assistant that delivers generative AI responses grounded in your organization’s own content.

It’s built natively into Qlik Cloud and designed around two primary components: the Assistant and the Knowledge Base.

This deep dive walks through how to configure, secure, and optimize Qlik Answers for internal use - whether you're supporting operations, legal teams, or customer service centres.

💡 Why It Matters

📂 80% of enterprise information is unstructured - PDFs, policies, presentations, emails, and more.

🔍 Traditional keyword search is brittle, and BI tools often ignore this data completely.

Qlik Answers bridges that gap. It gives users a natural language interface to search approved documentation with AI assistance - returning direct answers with source citations, not just a list of results.

By combining the recent upgrade to Claude Sonnet (via AWS Bedrock) and Retrieval-Augmented Generation (RAG), Qlik Answers delivers relevant, explainable outputs - all governed under Qlik Cloud’s enterprise-grade architecture.

💳 Licensing and Access

Qlik Answers is available as part of Qlik Cloud Analytics Standard, Premium, and Enterprise SaaS subscriptions. It is not included in the Starter tier.

Access to Qlik Answers is governed by capacity limits:

• Standard: 200 questions/month, 5,000 pages indexed

• Premium: 1,000 questions/month, 6,000 pages indexed

• Enterprise: 5,000 questions/month, 20,000 pages indexed

These limits apply to generative AI capabilities across your tenant. If needed, additional capacity can be purchased.

📝 Note: Licensing terms and limits may evolve. Always refer to the Qlik Pricing Page or consult your Qlik account representative for current details.

🧱 Key Concepts: Assistant and Knowledge Base

To deploy Qlik Answers, you configure two connected components:

• Knowledge Base: A curated set of content (documents, webpages, glossaries) that gets indexed for AI-assisted retrieval.

• Assistant: The user-facing chat interface. Each assistant can be linked to one or more knowledge bases.

Assistants are how users interact with Qlik Answers. Knowledge Bases are where the answers come from.

⚙️ How to Set It Up

1️⃣ Create a Knowledge Base

Navigate to Qlik Cloud Hub → Qlik Answers → Knowledge Base and click "Create knowledge base".

You’ll need to:

• Name the knowledge base and optionally add a description

• Choose where it lives (Private, Shared, or Managed space)

• Upload or connect your documents:

  • Supported formats: PDF, DOCX, XLSX, PPTX, HTML, TXT, MD

  • Connect to external sources like SharePoint, Dropbox, S3, Azure Storage

• Optionally tag or categorize the knowledge base for internal clarity

Once created, Qlik automatically parses, chunks, and indexes the content into a vector search backend.

💡 Tip: Keep each knowledge base tightly scoped. Avoid mixing HR policies with legal documents or product manuals in the same index.

2️⃣ Create an Assistant

Go to Qlik Answers → Assistants → Create Assistant.

You’ll walk through the following configuration steps:

• Name and description: This is what users will see in the assistant interface. Choose a clear, purpose-driven name like "HR Onboarding Assistant" or "IT Helpdesk FAQ".

• Linked knowledge base(s): Choose one or more previously created knowledge bases. You can link:

  • A single focused knowledge base (e.g. "HR Policies")

  • Multiple complementary sources (e.g. "Support Docs" + "Product Manuals") Each assistant only retrieves content from its linked knowledge bases. Keep the scope narrow to ensure relevance.

• User permissions: Assistants inherit access from the space they’re stored in. You can:

  • Store assistants in Managed Spaces for broad access

  • Use Private Spaces for early testing or internal-only assistants

  • Restrict access by assigning specific user groups to the space This ensures only authorized users can open the assistant and query the linked content.

• Conversation starters (optional but recommended): These are sample questions users can click when they first open the assistant. They help orient new users and set expectations.

  • Examples:

    • "What’s our policy on remote work?"

    • "How do I escalate a support ticket?"

    • "Where can I download the latest Qlik Sense user guide?" Add 3–5 questions tailored to the assistant’s domain. Keep them brief and informative.

• Embed settings (optional): You can generate an embed code to display the assistant in a third-party app, intranet portal, or web page. This allows users to engage with Qlik Answers directly inside their workflow.

You can also customize the assistant’s welcome message, which is shown to users at the start of a new conversation. This is useful for clarifying what the assistant is designed to help with.

Example: "Hi! I can help you find policies, guides, and answers about onboarding and HR procedures. Try asking: 'What’s the holiday policy?' or 'How do I access my payslip?'"

Additionally, you can configure error messages for common edge cases:

• Outside bounds of knowledge: When the assistant can't find relevant content

• Prompt injection detected: If the user attempts to manipulate the assistant’s behavior with unexpected input

• Too complex: If a query has too many layers and cannot be interpreted

Each message is editable, so you can tailor them to reflect your tone and audience.

Once saved, the assistant becomes available in Qlik Cloud as a live chat interface where users can begin asking questions.

🔐 Security and Access Scenarios

Qlik Answers respects Qlik Cloud’s space-based security model. That includes:

• Assistant access: Defined by where the assistant is stored (Private, Shared, or Managed space)

• Knowledge base access: Also space-dependent. If a user can't access a space, the documents in it are invisible to Qlik Answers

🧪 Scenarios:

1. User has access to assistant but not the knowledge base:

   • Result: Assistant loads, but won't return results from that knowledge base

2. User has access to multiple assistants, but only partial knowledge base access:

   • Result: Users only get responses from content they’re authorized to view

3. Admin stores an assistant in a managed space and links a knowledge base in a shared space:

   • Result: Both need aligned permissions for consistent experience

📊 Monitoring and Optimization

After launch, go to Qlik Answers → Review to:

• View all user questions and answers

• See upvotes/downvotes

• Identify unanswered queries or poor-quality responses

• Export logs to Qlik apps for deeper analysis

🔁 Use this data to iterate - add missing documents, adjust phrasing, or update outdated policies.

🛠️ Tips, Tricks & Advanced Scenarios

🧾 Document formatting: Use clear headings, short paragraphs, and structured FAQs where possible

📂 Assistant types:

• HR Assistant: Linked to onboarding docs, benefits policies

• Legal Assistant: Contract templates, definitions, compliance guidelines

• Support Assistant: Troubleshooting guides, escalation paths

🎯 Final Thoughts

Qlik Answers is not just an AI feature - it's an intelligent documentation layer.

With a well-structured assistant and targeted knowledge base, you turn unstructured data into uniquely searchable insight - without sacrificing security or trust.