Functions

Functions are custom Python modules that run in your private sandbox. They extend what your AI and workflows can do — from data processing and API integrations to custom calculations. You build them through conversation, and they become part of your system.

How it works

1

Describe what you need

Tell the AI what the function should do — in plain language. It generates the Python code, input/output schemas, and any dependencies.

2

Review and deploy

See a live preview of the generated code. Refine it through conversation or edit directly. When ready, it deploys to your private sandbox.

3

Use anywhere

Your function is now callable by the AI in conversations, as a step in workflows, or directly from the UI. Each execution is tracked with full history.

Function Builder

A conversational interface for building functions. Describe what you need, and the AI generates Python code with proper input/output schemas. You see a live preview of the code as it's being built and can iterate through conversation.

Screenshot placeholder

Isolated Execution

Every function runs in an isolated environment within your private sandbox — with its own dependencies and resource limits. Functions can't access each other's environments, and execution has configurable timeouts.

System Access

Functions can interact with your myOS system through a built-in SDK — execute capabilities, search your data, call the AI, and persist data across executions. They have the same access to your system that the AI does.

Execution History

Every execution is recorded — inputs, outputs, duration, and any errors. You can see exactly what happened on each run, making it easy to debug and iterate.

How functions are used

AI calls it

Your AI can invoke functions during conversations when it needs custom logic — data transformations, API integrations, or calculations that go beyond built-in capabilities.

Workflows use it

Functions can be steps in automated workflows — processing data, calling external services, or running custom business logic as part of a larger sequence.

Other functions call it

Functions can invoke other functions through the SDK, letting you build composable pieces of logic that work together.

Example Functions

Subscription Tracker

Scrapes your email for recurring payment receipts, extracts amounts and dates, and updates a tracking spreadsheet in your system.

Meeting Notes Processor

Takes raw meeting notes, extracts action items, creates tasks for each one, and links them to the relevant project.

Expense Categorizer

Reads transaction data from a CSV, categorizes each expense using AI, and creates a summary report with charts.

API Integration

Connects to an external API, fetches data on a schedule, transforms it, and stores the results as measurements in your system.

Code without the infrastructure

You don't need to set up servers, manage dependencies, or configure deployments. Describe what you need, and your AI handles the rest — from code generation to deployment in your private sandbox. Focus on what the function should do, not how to run it.