Technical

API, webhooks, performance, errors

• Is there API access?

  API access is available on Enterprise. You create your key in the panel settings and revoke it whenever you need to.

• What should I do if the system slows down?

  First check the status page: the website, data store and AI generation are measured there. If the problem continues, write to support.

• Is there downtime? What's the SLA?

  Unplanned downtime can happen. The availability target is written in our terms; we do not offer refunds or free months for downtime. The current state of each component is measured on the status page.

• What are the API rate limits?

  API calls are limited by an hourly quota per account. When the limit is exceeded the response is 429; the current limit is in the `X-RateLimit-Limit` header and the remaining calls in `X-RateLimit-Remaining`. Write to us if you need a higher quota.

• What happens if a webhook fails?

  AgentFabrica does not send outgoing webhooks to your systems at the moment, so there is no retry queue either. You can pull your data through the API.

• Can I bring my own AI API key?

  No. Generation runs on AgentFabrica's own infrastructure and counts against your plan quota; using your own key is not offered at the moment.

• Can I access operation logs?

  The team page shows the history of actions such as sign-ins, brand settings and team changes. System logs are not open in the customer panel; if you have a question about an action, write to support.

• How does AI Traffic work?

  We ingest pixel + GA4 + Shopify checkout events in real time. Hourly clustering + 7-persona classification (browser / considering / cart-ready / abandoner / customer / loyal / dormant). Intent score ≥85 + cart abandon → auto-triggers the WhatsApp Closer. Persona segments are pushed to Meta Custom Audiences hourly.

• How long does it take to turn a large catalog into ad creatives?

  It depends on the number of products and formats you pick. We do not promise a measured duration. Out-of-stock products are skipped during production. Live ads are not paused automatically when stock runs out.

• Does AI Mannequin look like a real model?

  From a flat product photo, it produces 6 variants by body type / age / ethnicity using diffusion + control models. No photoshoot, no agency — on-model imagery in minutes for fashion brands. Use it on /app/ai-mannequin. The background engine also drops seasonal scenes (summer/winter/holiday/Black Friday) behind the product.

• How do I trust suppliers in the Supplier Network?

  The Supplier Network has no registered suppliers yet; we do not show sample or rated profiles. For a product, /app/product-pipeline (Product Pipeline) shows the number of candidate suppliers and the cost range. Whichever supplier you work with, you build the trust yourself: ask for a sample before ordering, get delivery and return terms in writing and start with a small trial order.

• Which products does Auto Store Creator choose?

  Pick a niche → top-scoring products from Hand-Picked Winners + Niche Spy data are selected for that niche. They're added to your connected Shopify store as drafts (you approve and activate). Tag: `agentfabrica-auto-store`. Phase 2 will auto-push the brand kit (logo, colors, fonts) into Shopify shop settings.

• When an order comes in, is it auto-routed to the supplier?

  Yes — Shopify `orders/create` webhook → `splitOrderBySupplier()` → it determines which supplier ships each product → places the order via the dropshipping API → writes the tracking number back to Shopify fulfillment when received. Track each job's supplier mapping on /app/orders. Approval-first protocol: high-value orders require Telegram approval.

• Can I change the mascot's voice?

  No. The mascot's voice is shared by all accounts and set by our team in the admin panel; it cannot be changed from the customer panel.