Product · AI insights
AI that turns merchant data into next steps
Generic storefront AI already writes product copy. Merchant Support Systems targets the harder problem: analyzing loyalty and inventory signals and recommending what owners and managers should do next.
Demo uses scripted analyses — no API key required. Live SpaceXAI / xAI wiring is planned server-side.
AI demo
Domain analysis · scripted preview
AI analysis
DemoHealthy loyalty momentum with uneven inventory
Member growth is outpacing peers, while a small set of offline SKUs is dragging stock turns. Prioritize hybrid bestsellers and rebalance the oak desk lamp.
Insights
- Opportunity · Enrollment is a strength
- Risk · Offline SKU early risk
Actions
- High · Feature hybrid bestsellers
- Med · Review oak lamp merchandising
Structured analysis, not chat noise
Insights and ranked actions with severity and priority — designed for merchant ops, not open-ended chat that forgets your goals mid-thread.
Domain-aware: dashboard, products, loyalty, trends
Each area of the app can surface its own analysis shape. Today that runs on scripted demo data; live models come with real integrations.
Built for owners and managers
Actions are phrased for people who run the store — what to fix, who to re-engage, which SKUs to watch — not only for marketing specialists.
Loyalty
Program health
Inventory trends
Growth & risk
AI insights
Ranked actions
Merchant Support Systems
Owner · Manager · Online, offline, or both
The analysis shape
Reusable components in the codebase ( AiAnalysisPanel, insights,
actions) enforce a consistent structure merchants can learn once.
Insights
Observations with severity (e.g. risk vs opportunity). Explain what is happening in plain language.
Actions
Prioritized next steps with clear ownership potential — something a manager can execute this week.
Domain context
Analysis is scoped (loyalty vs trends vs products) so recommendations stay relevant to the page you are on.
Domains covered
Dashboard
Cross-cutting snapshot: which area of the business needs attention first.
Products
Channel mix, assortment issues, and catalog hygiene that affect sell-through.
Loyalty
Earn/burn balance, dormant members, and program design friction.
Trends
Climbers, stallers, risk SKUs, and what to restock or promote.
Alongside native storefront AI
Shopify Magic / Sidekick (and peers)
Strong at content, admin assistance, and in-platform shortcuts. Use them for merchandising speed.
MSS AI insights
Strong at cross-cutting operational judgment: program health, inventory growth patterns, and ranked actions for owner/manager workflows.
Product principles
- Prefer structured outputs over freeform walls of text
- Always show why an action is suggested (tie to an insight)
- Keep keys and model calls server-side when live AI ships
- Never pretend demo copy is live inference — label capability honestly
- Optimize for small teams: fewer, better recommendations
Frequently asked questions
- Is AI live today?
- No model calls yet. /app/ai-demo uses scripted analyses so you can evaluate UI, information architecture, and trust patterns. Live SpaceXAI / xAI (OpenAI-compatible API) is planned with server-side keys only.
- Will AI replace my manager?
- No. It ranks suggestions. Humans still know local context, supplier quirks, and customer relationships. The goal is fewer blank pages at the start of the week.
- How do you prevent hallucinations?
- Live design will ground outputs in merchant data, constrain schemas (insights + actions), and keep humans in the loop. Demo content is authored, not improvised, so it is stable for walkthroughs.
- Can I turn AI off?
- Product direction: AI is an assistive layer on top of core loyalty and trends views, not a hard dependency for using the app. Controls will land with live features.
- Where do I try it?
- Open /app/ai-demo for the interactive panel across domains. Marketing page you are on explains the capability for stakeholders who will not log into the app yet.
Try the AI analysis demo
Switch domains, inspect insights and actions, and share the flow with stakeholders — no API key needed.