> ## Documentation Index > Fetch the complete documentation index at: https://www.twicecommerce.com/docs/llms.txt > Use this file to discover all available pages before exploring further. # The Operating System approach > We believe modern commerce doesn't need more vertical tools — it needs horizontal, configurable layers that connect data, workflows, and systems. In a world moving toward agentic commerce and AI-driven automation, clear semantic structures and well-defined ontologies matter most — they make data understandable, accessible, and ready for intelligent action both by humans and AI. In practice, it means you organize data and views the way your business already runs — not the other way around. We make sure that the configuration is accessible by other key service providers and AI Agents - and in our case that it fits the needs of recommerce. We bring this belief to life through a few key platform decisions listed below. Each of these decisions reflects how we think modern commerce systems should work — flexible, connected, and data-driven — while also addressing the *unique needs of recommerce*. The result is that TWICE can stand alone as a complete commerce platform with best-in-class recommerce capabilities, or plug seamlessly into your existing stack as a *recommerce operating layer*. ## 1. Core assets with lifecycle support Commerce runs on four essential assets — **inventory** (what you have), **catalog** (how you present it), **orders** (what’s moving), and **customers** (who it’s for). In TWICE, these aren’t isolated modules but *connected data models with full lifecycle tracking*. For recommerce, that lifecycle doesn’t end at the point of sale — it often continues through *use, return, refurbishment, and resale*, capturing changes in *condition, value, and ownership* along the way. Orders both bring items in and send them out — helping you track how products move, change condition, and keep earning over time. ## 2. Table-first approach Most of TWICE's core views are built as [data tables](/docs/admin/tables) — because that's how operators actually work. Every table view is fully *searchable, filterable, sortable, and groupable*, with full *column controls* (pin, hide, reorder) and one-click *export*. You can build and save *custom views* (similar to Notion) to fit your daily workflows and roles. What makes these views so powerful, is the fact that you can *store any data about any asset* using custom [attributes](/docs/concepts/admin/attributes), instead of being locked into a predefined set of fields. Every business models its products and processes differently — so we let you define the data that matters to you and surface it directly in your views. Whether you’re a store clerk tracking active orders, a manager reviewing stock movement, or an owner monitoring performance, TWICE helps you *highlight what matters to you* — not what the system decides is important. For recommerce, tables natively extend to handle *condition, lifecycle stage, and item history* — letting teams manage circular inventory just as easily as new stock. This table-first logic keeps everything *consistent, transparent, and adaptable* across inventory, catalog, orders, and customers — whether you’re a *human, an integration, or an AI agent* working on top of the same data. ## 3. Configuration layer Every business operates differently — so TWICE lets you configure the system to match how you actually work. You can define *automation rules* through [workflows](/docs/workflows): when a trigger happens, the system can automatically *create, edit, or update assets* like orders, inventory items, or customers. It’s how you turn operational logic into repeatable automation. TWICE supports *multi-location and multi-tax operations*, with assets moving freely between warehouses and fulfillment types — from *click & collect* to *shipping and returns*, all running in parallel. We think of our OS like a *steam engine*: the settings you define determine how much power you generate. You can push your operations as far as you want — even to the point of overpressure (in commerce, that might mean selling when out of stock). We provide the engine; *it’s up to you to decide how close to the limit you want to run it*. For recommerce, this flexibility extends to *inspection, repair, restock, recycling, and spare part tracking* — recognizing that modern operators need to configure workflows beyond traditional sales and fulfillment. TWICE lets you define and manage these circular processes as part of your core operations, not as exceptions. ## 4. API & MCP layers TWICE is built *API-first* — every function you see in the Admin is powered by the same public API you can access. See the [API reference](https://server.twicecommerce.com/api/internal) for full endpoint documentation. This makes the platform *headless and composable* by design, allowing you to integrate TWICE with any system in your stack or build your own tools directly on top of it. You can *read, create, or update any asset* — inventory, catalog, orders, or customers — through standardized endpoints that are always available. This consistency keeps integrations predictable, whether you’re connecting to ERP, POS, logistics, or AI systems. The same layer powers our *Model Context Protocol (MCP)* — giving AI agents structured access to your commerce data and workflows, just like a human operator in the Admin. We invest heavily in making your operational semantics and ontology clear, accessible, and interpretable to the best AI agents on the market. For recommerce, this means agents can understand product lifecycles, conditions, and value states — and act intelligently across systems without breaking context. In short: TWICE exposes your entire commerce operation as a *connective layer*, ready for humans, integrations, and AI to work together on the same data. We also provide these docs (including guides and API references) through the [MCP layer](/docs/setup-mcp-access).