Skip to content

Start here

Welcome to Strive

Understand the workspace and how apps, workflows, connections, and data fit together.

Strive brings an interface, the information behind it, and the processes that keep it useful into one workspace. You can start with an app template, build an app of your own, or ask the Assistant to help you assemble one.

How the pieces fit

PieceWhat you use it forExample
AppA place to perform a taskReview invoices, follow news, or inspect documents
WorkflowThe steps that do the workRead a file, extract values, and prepare a result
ConnectionAccess to a service or storage locationGoogle Drive, a database, or managed storage
Data sourceA collection of records produced or read by a processExtracted invoice fields from an ingest step
RecordOne item of information, with its originA row, document segment, or search result
AssistantHelp across the workspace or on the current pageCreate an app, explain a failed run, or find a file

An app can contain several workflows and read several sources. A connection can be reused by more than one workflow. Running a workflow and opening an app are separate actions: opening an app can show existing results without processing everything again.

Choose your starting point

Use the first-app guide to get a working example. Read navigation to find your way around. If you already know the task you want to accomplish, start a conversation in the Assistant.

Your workspace and other people's apps

Your organisation supplies the workspace, connections, and available capabilities. Apps have an owner and can have their own sharing settings. Being able to open a shared app does not necessarily let you edit it, manage its connections, inspect its owner-only reviews, or open the rest of its owner's workspace.

Some controls depend on your role and the app's type. A template app can expose parameters; an app that does not declare parameters has no Parameters section. This manual explains each control where it applies.

A useful first session

Create a small app, run it on a few representative inputs, inspect the output and its source, and then decide who should use it. Test both an ordinary input and an empty or incomplete input before relying on an automated process.