How Bowties works

Everything you need to know to go from raw docs to agent-ready code — in one session.

Overview

Bowties is a developer tool that transforms your existing app documentation into two production-ready outputs simultaneously: Swift App Intents scaffolding for Xcode, and a structured JSON-LD schema optimized for Siri retrieval and MCP tool calling.

The core insight is simple: the same source material that describes what your app does can generate both the technical integration layer (App Intents) and the semantic retrieval layer (agent-readable content) in a single AI-powered pass.

💡

Why both outputs? App Intents tells iOS how to invoke your app from Siri. The JSON schema tells agents what your app can do and how to talk about it. You need both to be fully agent-ready after WWDC.

Quickstart

01

Pick a template

Choose from API Docs, Product Page, Help Center, or Custom based on what you're pasting.

02

Paste your content

Drop in API documentation, a product description, or help center text. Plain text works fine.

03

Click Generate

Both outputs arrive in parallel in under 30 seconds. Copy or download each with one click.

Keyboard shortcut

Press ⌘ + Enter (Mac) or Ctrl + Enter (Windows) to trigger generation from anywhere on the page.

Example input

Input — API Docs template 
Our app lets users search for products by category, add items
to a cart, and complete checkout.

Endpoints:
  GET  /products?category=shoes&limit=20
  POST /cart/add  { item_id: string, quantity: number }
  POST /checkout  { payment_token: string }

The app also supports user authentication via GET /me
and order history via GET /orders.

Input templates

Each template uses a different system prompt tuned for its source type. Choosing the right template improves output quality significantly.

API Docs

API Documentation

Best for REST API references, endpoint lists, or OpenAPI specs. Optimizes for parameter extraction and action mapping.

Product Page

Product Page

Best for feature descriptions, marketing copy, or app store descriptions. Focuses on top user actions Siri would surface.

Help Center

Help Center

Best for FAQ content, support articles, or how-to guides. Extracts Q&A pairs and task-based intents automatically.

Custom

Custom

For anything else. Bowties uses its best judgment to identify key intents and structure the outputs appropriately.

Swift App Intents output

The Swift output is production-ready scaffolding following Apple's App Intents framework (iOS 16+). It's designed to be dropped directly into an Xcode project with minimal editing.

What's included

  • Correct @AppIntent protocol conformance for each identified action
  • Entity definitions with @Entity for core data types
  • AppShortcutsProvider with natural language Siri phrase strings
  • @Parameter declarations with correct Swift types
  • perform() methods with realistic IntentResult returns
  • Inline comments explaining each section
Example output — Swift App Intents 
import AppIntents
import Foundation

// MARK: - Search Products Intent
struct SearchProductsIntent: AppIntent {
    static var title: LocalizedStringResource = "Search products"
    static var description = IntentDescription("Find products by category")

    @Parameter(title: "Category")
    var category: String

    func perform() async throws -> some IntentResult {
        // TODO: call your API here
        return .result(value: "Found products in \(category)")
    }
}
⚠️

Generated code is scaffolding, not final implementation. You'll need to replace // TODO sections with your actual API calls and business logic.

Agent-Ready JSON output

The JSON output is a structured schema document that makes your app's capabilities retrievable and actionable by AI agents, including Siri with Apple Intelligence and any MCP-compatible system.

Schema structure

  • mcpTools — MCP-compatible tool definitions with inputSchema (JSON Schema format) for each action
  • appCapabilities — Natural language strings describing what your app can do, for agent context
  • siriFAQ — Question/answer pairs for likely Siri queries, structured for retrieval
  • entities — Data model definitions for your app's core types
  • @context — schema.org JSON-LD context for semantic interoperability
✅

The JSON output is designed to be compatible with MCP server tool definitions. Once Apple ships MCP support at WWDC, this schema can be used directly with the native integration layer.

Why WWDC 2026

Apple is expected to announce native agentic infrastructure for iOS at WWDC 2026. Based on developer leaks and analyst reports, this includes App Intents becoming the primary interface for Siri to interact with third-party apps, native MCP integration at the OS level, and ambient intelligence across the phone available to any app that adopts the framework.

The apps that ship App Intents support in the first 90 days post-WWDC will benefit from increased Siri visibility, featuring in Apple's WWDC demo apps, and first-mover advantage in agent-driven discovery. Bowties exists to compress that prep time from weeks to minutes.

MCP + Siri explained

MCP (Model Context Protocol) is an open standard for connecting AI models to tools and data sources. It defines how an agent should describe available actions (tools), what parameters they take, and how to invoke them.

Bowties generates MCP-compatible tool definitions in the JSON output. Each tool definition follows the JSON Schema format that MCP servers expect, making the output immediately usable in any MCP-aware system — not just Apple's ecosystem.

Siri with Apple Intelligence uses a similar retrieval pattern: it needs to know what your app can do, in structured form, to surface it proactively. The siriFAQ and appCapabilities fields in the JSON output are designed to feed that retrieval layer directly.