The Omnipresent AI Chat Assistant is a conversational AI embedded throughout Ardoq that answers questions about your architecture in plain language. Building on the original Reports-focused Chat Assistant, this expanded release brings natural language querying to every corner of the platform — Dashboards, Viewpoints, the Metamodel, and Reports — so anyone can access architecture intelligence without having to have pre-made assets or the ability to run advanced queries.
Availability: General Availability (Q2 2026). Requires Ardoq AI features to be enabled for your organization. Admins can activate the feature in Organization Settings → Features.
What's New in the Omnipresent Assistant
The original AI Chat Assistant was scoped to Reports only. The Omnipresent AI Assistant expands coverage to the full platform
Full platform reach. Works across Reports, Dashboards, Viewpoints, and the Metamodel.
Improved accuracy. Extended thinking significantly improves accuracy on quantitative questions, counts, and comparisons.
Context aware. Prioritizes the asset you're currently viewing — such as a report or dashboard — to answer questions first, then falls back to other sources.
Truly omnipresent. The assistant follows you as you navigate Ardoq, always available wherever you are.
What Does It Do?
The Omnipresent AI Assistant lets you interact with your Ardoq data using natural language. Below are the primary capabilities:
Query the Architecture in Plain English
Ask questions about Components, Viewpoints, and the Metamodel without opening assets or having to write GREMLIN queries. The assistant routes your question to the right data source automatically.
Example: "Which applications connect to our SAP system?"
Example: "What does our Metamodel say about the relationship between Business Capabilities and Applications?"
Example: "Show me the technology stack for our customer-facing services."
Follow you around in Ardoq and adapt to your context
The assistant is context aware — it prioritizes the asset you're currently viewing, whether that's a report, dashboard, or Viewpoint, and uses it to answer your questions first. As you navigate Ardoq, the assistant follows along and adapts automatically, so you never need to tell it where you are or what you're looking at.
Example: When viewing an Application report - "Show only applications with a criticality of High."
Explore and Summarize Reports
Summarize report content, perform quantitative analysis, identify risks and action points, and explain findings to non-technical stakeholders — all through conversation.
Example: "What should I look at first here?"
Example: "Explain what this report means for the Finance team."
Reason about Viewpoints and traverse the Graph
Ask questions about what a Viewpoint shows and use it as a starting point to explore related components, references, and dependencies across the graph. The assistant uses the Viewpoint's context to traverse connected data and surface insights you might not find in a flat list.
Example: "Walk me through the dependencies shown in this Viewpoint."
Example: "How are the components in this Viewpoint connected to our core banking platform?"
Understand Dashboards
Ask the assistant to interpret charts, trends, and underlying metrics shown in any dashboard. CIOs and senior stakeholders can make architecture-informed decisions without waiting for an EA to translate the data.
Example: "What's driving this spike in critical applications?"
Example: "Which components are most at risk based on this dashboard?"
Who Can Use It?
All organizations with Ardoq AI features enabled. An Admin must activate the feature in Organization Settings → Features.
Any user with permission to view the underlying data. Responses are permission-aware and will never expose content the user is not authorized to see.
Contributor users accessing Ardoq via the New Discover Experience can also use the assistant.
Primary audience: Enterprise Architects, CIOs, business analysts and cross-functional stakeholders who need architecture insights without technical query skills.'
How to Access It
The assistant is available everywhere, and it will be open by default when you log in to Ardoq - you do not need to open a specific view to start a conversation. If you close the assistant, just look for the AI assistant icon to reopen it — it appears in the toolbar across all locations in Ardoq.
To start a conversation:
Open any location in Ardoq
Click the AI assistant icon in the top-right corner of the panel.
Type your question in plain language and press Enter.
Use follow-up questions to drill deeper — the assistant keeps full conversation context.
Rate responses with thumbs up or thumbs down to help improve quality.
No setup is required. The assistant is native to Ardoq.
Features
Reasoning Chain
Every Chat response contains a Reasoning chain in addition to the response. The Reasoning is collapsed by default, but can be expanded to examine the AI's thought process in generating the current answer. Use it to understand how the AI Assistant arrived at its answer, and to check that the reasoning is sound.
Chat History
Conversations are saved so you can return to previous sessions and continue where you left off.
Accessing Chat History
To access the chat history, click the history button on the top left corner of the Chat Assistant window.
You will see all conversations independent of the user’s location in Ardoq when the question was answered. When users open an historic conversation, they stay on the same page / location in Ardoq and are not automatically navigated anywhere.
Conversation History Search
Search the history overview by keyword to quickly find a previous conversation.
Pin and Delete Conversations
Hover over a conversation, click the three-dot icon, and select Pin or Delete. Pinned conversations appear at the top of the list. Note: deleting a conversation is permanent.
Assistant modes
The AI assistant can be opened in several view modes - sidebar, floating and full screen. To select mode click the mode icon in top right corner of the assistant window.
Sidebar mode
The default mode of the assistant, it sits in a dedicated panel on the side of the Ardoq application, allowing you to view and interact with the architecture content and the assistant simultaneously. You can resize the sidebar
Floating mode
In this mode, the assistant appears as a window that overlays the main Ardoq content. This is useful for keeping a conversation open while navigating different parts of the application or focusing on a specific piece of architecture data.
Full Screen mode
This mode maximizes the assistant window to cover the entire viewport, except for the main Ardoq sidebar menu. Full Screen mode is ideal for deep dives, extended research, or when focusing entirely on a complex Q&A session without distraction from the underlying architecture views.
AI Assistant for Business user (Discover)
Give business users instant, plain-English answers about your architecture from the Discover landing page. No clicking through menus, no waiting on an EA to pull a report; the AI assistant just provides them direct answers based on the data and assets shared with them.
What it looks like
On the Discover landing page, the standard search bar is replaced with "Ask Ardoq AI a question…". AI chat input with a suggestion chip below it.
Note: Search is still available via Cmd/Ctrl+K or from the sidebar. Your theming and the rest of the home page are unchanged.
How to use it
To start a conversation from Discover:
Type your question in the AI chat input, or click one of the suggestion chips.
Press Enter. The assistant opens in Full Screen mode with your question already sent.
Use follow-up questions to drill deeper. The assistant keeps full conversation context.
To return to Discover, use the back button. Your conversation stays in chat history and can be picked up later.
What needs to be enabled
The AI assistant for business users is only available if you have also opted into the New Discover Experience. The AI assistant won’t be available to users through Legacy Discover.
You can enable it for your organization by going to:
Organization Settings → Features:
New Discover Experience
Omnipresent AI Assistant.
Tips for Great Results
Be specific. Include the scope or angle you care about, for example 'for EMEA' or 'applications owned by the Cloud team.'
Use follow-ups. The assistant keeps your conversation context, so ask follow-ups to go deeper.
Check the source. If something looks off, check the underlying view to validate. The assistant summarizes what is already in Ardoq.
Stay architecture-focused. The assistant is optimized for your Ardoq data — it works best when questions are grounded in your architecture, not general knowledge questions.
Use focused reports. For reports, smaller focused reports yield more accurate results than very large ones.
Use Ardoq assets for best results. Create assets, such as Reports, Dashboards and Viewpoints to answer common user questions. Without assets, the assistant uses less reliable methods to finding answers. Remeber to give users access to the assets (including Contributors, if relevant)
What Is Included
Natural language querying across Reports, Dashboards, Viewpoints, and the Metamodel
Summaries, interpretations, and quantitative analysis
Multi-turn chat with full conversation context
Saved chat history with search, pin, and delete
Reasoning chain transparency on every response
Aggregation tools for improved accuracy on counting and calculations
Permission-aware responses aligned to each user's access rights
Known Limitations
AI can make mistakes. The assistant uses a large language model and can produce errors, especially on complex multi-step calculations or edge cases. Always verify critical outputs against the underlying data.
Very large datasets. Very large datasets — for example, reports with thousands of rows or Deep-spanning Viewpoints — may produce slower or less reliable responses.
Access control. The assistant can't reason on information it can't access. So if non-admins with limited access use the assistant, their answers may differ because of access control.
Non-English queries. Quality of responses in non-English languages may vary. The assistant supports many languages, and will usually respond in the same language as the query, but full support of all languages is not guaranteed.
Security and Privacy
Responses are permission-aware and aligned to what each user is authorized to see. Users will never receive data outside their existing access rights.
The assistant is built natively into Ardoq Core. No external setup or data export is required.
For general information about security and privacy in Ardoq AI features, refer to the Ardoq AI Privacy FAQ.
Responsible Use Guidelines
Always verify results. AI responses are suggestions for accelerating analysis and can be inaccurate or incomplete. Always verify important findings against the source data before acting on them.
Do not input sensitive data. Do not submit personally identifiable information (PII), proprietary passwords, or confidential information into the chat interface.
Architecture-focused use. The assistant is purpose-built for Ardoq architecture data. Use it for architecture-related questions rather than general knowledge queries.
Frequently Asked Questions
Q: Does the AI Assistant change my data?
A: No. The assistant only reads and interprets data that already exists in Ardoq. It does not create, edit, or delete any records.
Q: Can it search across all workspaces and artifacts?
A: Yes — in Q2 2026 the assistant is expanded to cover Reports, Dashboards, Viewpoints, and the Metamodel across Ardoq.
Q: How do I get access?
A: Your Ardoq organization must have AI features enabled. Once enabled, an Admin can activate the Omnipresent AI Assistant in Organization Settings → Features.
Q: Is the AI Assistant available in Discover?
A: Yes, but only in the new Discover experience. If your organization has upgraded to the new Discover experience and has the Omnipresent AI Assistant enabled, the assistant will be available to Contributor users automatically. Learn how to enable the new Discover experience.
Q: Can I ask questions in my native language?
A: Yes. The underlying AI model supports many languages and will respond in the language of your query. Response quality may be lower for non-English languages.
Q: What happened to the Reports-only Chat Assistant?
A: The Omnipresent AI Assistant supersedes the original Reports Chat Assistant. All existing functionality — multi-turn chat, chat history, reasoning chains — is preserved and extended to cover the full platform.
Q: Is the assistant available to Contributor users?
A: Yes. Contributor users accessing Ardoq via the Discover Experience can use the assistant, subject to their existing data permissions.
Q: How is this different from a generic AI copilot?
A: Unlike general AI copilots layered onto platform data, the Ardoq assistant is purpose-built for enterprise architecture data structures. It understands Ardoq constructs — Components, Viewpoints, Metamodel — natively, uses a neuro-symbolic approach combining AI, rules, and graph reasoning, and is permission-aware by design.
How to Give Feedback
Rate any response using the thumbs up or thumbs down icons in the chat window. Your feedback directly improves the quality of future responses.
Thumbs Up: Use this to signal that an answer was helpful or accurate.
Thumbs Down: Click thumbs down if an answer is not quite right. You will be prompted to categorize the issue (Inaccurate, Incomplete, Error, or Other) and optionally add detail.