Skip to main content

Workflows

Workflows are where you configure your customer's use cases. This is the heart and soul of how the AI behaves — whatever you define here controls which flow is triggered, and what happens inside that flow.

The fastest way to build and change workflows is the Copilot. This page describes the building blocks.

Flows​

Under Workflows → Flows you'll find the flows that have been configured. A flow is a unit of AI behavior — for example, handling "where is my order," processing a return, qualifying a lead, or answering from your knowledge base.

You can add and edit flows through a form, but in practice the team rarely uses the form directly — most work is done conversationally through the Copilot. The Copilot can read every configured flow, explain how a journey is handled, and make changes for you.

A flow typically defines:

  • Objectives the AI should complete.
  • Questions the AI should ask and the answers it expects.
  • Data points to extract from the conversation or look up.
  • Custom API / integration lookups for real data.
  • Ticket fields to capture and child tickets to create.
  • Completion and handoff rules — when to close, and when to hand to a human.

For the full configuration reference — every objective kind, conditions, transitions, and completion behavior, and how each shapes the conversation — see Flows.

Testing a Flow​

The workflow editor includes a Test flow feature so you can try a flow before relying on it. You can hold a manual conversation with the AI, or describe the conversation you want and let the AI run it for you (for example, "create a conversation where you provide your order ID"). For repeatable, assertion-based testing, see Testing.

Data Library​

The Data Library is where workflows get the data they use to answer customers — order details, customer details, order lists, custom API responses, Google Sheets data, and standalone data points. For the built-in order and customer data points, custom-API-backed lookups, and execute actions in detail, see Data Points.

It also includes an Objective Library: objectives you want to reuse across flows. Mark an objective as shared and it becomes available to import into other flows, so common logic is defined once.

Sample Data (Response Library)​

For the AI to be configured well, it needs realistic API responses to work against — this is called sample data.

For example, to build a "delayed order" workflow you first need an actual delayed order to design against. You fetch a delayed order (by customer ID or order ID), save its response as, say, "delayed order," and the AI uses that saved sample to configure the workflow.

You can save sample data manually, or just ask the Copilot — e.g. "save this order ID response as delayed order."

Products​

The Products page lists the products you've configured or synced. Products give the AI catalog context for product search and recommendations. (Products have a lot of depth; this is the high-level view.)

Stores​

Configure your offline stores here so the AI can answer location questions — for example, "where is the nearest store?"

Knowledge Base​

The AI also answers from a question-and-answer knowledge base (FAQs). Unlike open-ended document upload with retrieval, Flowcall expects knowledge in a structured Q&A format, which makes answers more reliable. See Knowledge Base (FAQs).

  • Flows — the full flow configuration reference.
  • Data Points — built-in and custom-API-backed data points and execute actions.
  • Custom APIs — the endpoints workflows call, and source APIs for built-in data.
  • Copilot — build and change workflows by chatting.
  • Testing — automated, repeatable tests with assertions and mocking.
  • Integrations → API Library — connect the custom APIs that workflows call.