The Evaluations framework in Xcode 27 lets developers generate and validate synthetic test data at scale using the SampleGenerator API, then run robust evaluations against agentic workflows that involve tool calling ā all from Swift code.
ā¢ Eliminates hand-writing hundreds of test cases: the makeSamples / SampleGenerator API generates diverse synthetic data programmatically, exposing gaps a small handcrafted dataset would miss.
ā¢ Built-in validation closures let you enforce structural rules (tag count, string length, casing) per sample at generation time, keeping bad data out of your evaluation set automatically.
ā¢ Supports agentic / tool-calling workflows specifically, so you can evaluate multi-step AI features ā not just single prompt-response pairs ā and track quality over time with the Xcode 27 Evaluations Report.
Foundation Models is a new Apple framework introduced in iOS 27 that gives developers on-device access to the same Apple Intelligence language model powering system features, enabling text generation, structured output, and tool-calling entirely on-device without a network connection.
iOS 27 opens the Foundation Models framework to third-party LLM providers via a new public LanguageModel protocol, enabling anyone to integrate custom, server-based, or open-source models using the same Swift API as Apple's on-device system model.
App Schemas let developers describe their app's content and actions using pre-defined domain schemas (like the Calendar domain) so Siri can understand, search, and act on app data without custom NLP. Entities conforming to IndexedEntity are donated to Spotlight's semantic index, enabling natural-language queries over app content.
In-depth guide
iOS 27 On-Device AI & Apple Intelligence āVisual Intelligence brings iOS 17's Visual Look Up capabilities to a new developer-facing API surface in iOS 27, letting apps pipe live camera frames or static images through on-device scene understanding to extract subjects, text, barcodes, and rich semantic labels without any cloud round-trip.