Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.vortexiq.ai/llms.txt

Use this file to discover all available pages before exploring further.

Ask Viq is the conversational interface of Vortex IQ. You type any question about your store in plain English — “which campaigns are losing money this week?”, “why did refunds spike on Tuesday?”, “what’s my margin by SKU after landed costs?” — and Ask Viq routes the question against your connected sources, finds the relevant data, and returns an answer with cited sources you can click into. Every answer traces back to a real data point. Nothing is invented.
The brand name is always written as two words: Ask Viq. Never abbreviate it to “Viq” alone. This applies everywhere — in documentation, in product copy, and in any generated prose.

How Ask Viq differs from a chatbot

Most ecommerce chat tools wrap a language model around one platform’s data and call it AI. Ask Viq is different because it sits at the centre of the Vortex IQ AI OS and reads from the entire connected graph: 215 connectors, 6,034 KPI cards, every Vortex Mind report archived in Vortex Memory, and the structured findings the audit engine has produced over time. When you ask Ask Viq a question, it does not guess. It routes your question against the four bodies of data that make up the AI OS substrate, composes an answer from what it finds, and attaches a citation to every factual claim. You can click any citation and land on the underlying source — the same live data your dashboard shows.

What you can query

Ask Viq draws answers from four bodies of data, all filtered by your active Profile.

Nerve Centre cards

6,034 live KPI cards across 215 connectors. Every metric the active Profile has a connected source for is reachable — total revenue, auth rate, refund rate, ROAS by campaign, and thousands more. Card definitions are stable and reconcile to the vendor’s own dashboard.

Vortex Mind reports

The structured findings from every Vortex Mind diagnostic run: cause, evidence, severity, and suggested action. All seven OpenClaw report types write findings into the same archive, and Ask Viq reads from all of them without you needing to know which report produced which finding.

Vortex Memory

Knowledge files you have uploaded (PDFs, spreadsheets, policy documents, supplier catalogues) plus every archived Vortex Mind report run — giving Ask Viq longitudinal context that deepens the longer you use the platform.

The connector graph

Cross-connector joins that no single tool in your stack can produce alone: ad spend against inventory state by SKU, Stripe declines against order history by customer, Klaviyo flow recipients against the order index by date.

Hallucination guardrails and citations

Ask Viq is engineered around three layers of guardrails that make hallucination structurally unlikely rather than statistically rare.
Every factual claim in an Ask Viq answer carries a citation. Numbers come from named Nerve Centre cards. Diagnostic narratives come from named Vortex Mind report runs. Reference statements come from named Vortex Memory files. You can click any citation and land on the underlying source in two clicks. Citation enforcement is structural — the model cannot emit a number that did not originate in a cited source.
1

Layer 1 — Required citations

Every numerical claim must carry a citation token pointing to a specific card, finding, or file. If the model generates a number it cannot trace to a cited source, the answer is rejected and regenerated before it reaches you. This layer is not configurable — it is a structural property of the AI OS.
2

Layer 2 — Workspace AI Guardrails

Your workspace admin can maintain a forbidden-words list on the Settings, Brand and AI tab. Every Ask Viq generation is filtered against it. Brand-protection language, regulatory claims, and competitor names you want excluded never surface in generated prose — regardless of how a question is phrased or which Brand Voice Profile is active.
3

Layer 3 — Model-level prohibitions

Ask Viq will not project beyond available data, will not opine on plan tier or pricing, will not invent business context that has no uploaded file behind it, and will not produce statements about data outside your active Profile. These rules apply universally and cannot be bypassed through prompt engineering.
When any layer blocks an answer, Ask Viq returns a clear, polite limit message rather than a fabricated response. You always know when data is unavailable and why.

Query history

Ask Viq is not a stateless chat. Every conversation persists as a named session in the left sidebar, grouped by date band (“Today”, “Yesterday”, “Apr 2026”, “Mar 2026”). You can pin sessions you return to often, search across every session you have ever had, and export any session in three formats.
Every session records the question text exactly as typed, the Profile that was active, the user who asked, the full structured response including all citations, the “as of” timestamps from each cited source, and the pin and archive state. History is retained indefinitely by default; workspace admins can configure a retention policy under workspace admin settings.
Within a single session, Ask Viq carries context from prior turns. You do not need to restate the time window or entity on every follow-up. “Now break that down by region” works because the prior turn’s metric and window carry forward automatically. Across sessions, context resets unless you pin and reuse a session.
The search box at the top of the sidebar searches across every session you have access to — matching question text, answer text, session title, and citation source names. A search for “refund_rate” surfaces every session that ever cited that card. Search respects Profile and permission boundaries.
Any session can be exported as HTML (full thread with citations, formatted for sharing), Markdown (prose only, for pasting into a document), or JSON (structured form with question, answer, citations, profile, and timestamps, for programmatic processing or audit review).
The longer your workspace uses the platform, the more material accumulates in Vortex Memory. After a year, multiple seasonal cycles are captured as archived report runs. Ask Viq can reach into past Novembers to answer “why does refund rate always spike in late November?” and surface recurrence detection — “this looks similar to the issue you saw on 03 Mar 26, here is what was done about it then.” That institutional memory survives personnel turnover.

How Ask Viq connects to other modules

ModuleWhat Ask Viq draws from it
Nerve CentreLive KPI cards — the primary substrate for current-value questions
Vortex MindStructured diagnostic findings — the primary substrate for “why” questions
Vortex MemoryKnowledge files and archived report runs — longitudinal context and reference documents
Settings, ProfilesThe Profile registry that the Profile dropdown reads live
Settings, Brand and AIGuardrails configuration and Brand Voice Profile shape