Core concepts

Flexkit brings three things together: your store’s data, the workspace your team uses, and an AI agent that works with both. This page explains the pieces you meet in the app.

Teams and projects

A team is your company or working group. It owns projects, pays for them and holds its members.

A project is one business, or one store. It holds the data model, the records, files and documents, the people who can work on it, the agent’s memory and skills, and the automations. Most companies need one project; separate projects make sense for businesses that share nothing, or for a test copy of your setup.

Studio: the app you work in

Studio is where you do the work, in the desktop app or in your browser at flexkit.app . Its apps sit in the rail on the left:

App	What it is for
AI	Chat with the agent, automations, plugins, skills and approvals.
Desk	Browse, create and edit records, such as products, orders and customers.
Asset Manager	Upload and organize images and other files.
Knowledge	Documents the agent uses as sources: manuals, contracts, price lists.
Graph	See how records connect.
Explorer	Run GraphQL queries against the project, for technical users.

The Dashboard is the administration side: members and roles, plans, usage and billing.

The schema: your data model

The schema describes what your business is made of. An entity is a kind of thing, such as Product, Order or Supplier. Each entity has attributes (fields), such as a name, a price or a status. Relationships link records together: an order belongs to a customer, a product has variants. A record is one saved item, such as one product.

Desk’s lists and forms, the agent’s understanding of your business and the API all come from the schema, so it decides what questions the agent can answer. You usually build it with the agent: start from the commerce starter or a description of your business, then ask for changes as you go; the agent shows each change for approval before applying it. Developers can manage it in code instead. See schema.

Two more ideas come up as a project grows:

A scope is a variant of the data, such as a market or a language. A product can have a different name and price in the UK and in the US while sharing its SKU.
A space limits who can see something. Records, fields and documents in a space are visible only to its members.

The Flexkit Agent

The agent works in the chat, in automations, and wherever you connect it, such as Slack. It reads your records and documents, runs queries, writes reports and exports, and uses connected services. It always acts with the access of the person it works for: it cannot see records hidden from you.

It does not change anything on its own. Changes to records, the schema or connected services are shown to you as an approval first; you approve or reject them. You can let a trusted automation apply certain changes without asking.

You shape how it works with:

Skills: written instructions for a task, such as how you price a product or answer a return request.
Memory: what the agent learns about you and the project over time.
Plugins: connections to services such as Gmail, Shopify, Notion or Google Sheets.
Custom tools: your own code that the agent can call, written by a developer.

Automations

An automation is a job the agent does on its own: a set of instructions with a trigger. It can run on a schedule, when records change, when another system calls a webhook, or when you start it. Each execution is a run you can inspect afterwards in Run History. The agent can set up an automation for you from the chat. See automations.

For developers

The project’s GraphQL API is generated from the schema, so storefronts, scripts and other systems use the same data the agent does; see APIs. The CLI covers projects, schema, data and the agent from a terminal. A project’s schema can be managed from code in flexkit.config.ts and deployed with flexkit deploy, and teams that need the Studio inside their own application can self-host it.

Next: create a project or join your team.