Apple Reaffirms Revamped Siri Launch for 2026

- The revamped Siri is built on a new architecture, internally codenamed "Linwood," which utilizes Apple's own large language models (known as Apple Foundation Models) and also incorporates technology from Google's Gemini team for certain capabilities. - A core goal of the overhaul is to introduce "agentic" AI that can perform complex, multi-step tasks across different applications, a shift from simply responding to queries to autonomously executing workflows. - The new feature suite, branded "Apple Intelligence," will process many tasks on-device for privacy and speed, using an algorithm to determine when to leverage more powerful cloud-based models. - Development has faced hurdles, with internal testing revealing issues where Siri would incorrectly default to its OpenAI ChatGPT integration instead of using Apple's own models for tasks it should handle. - The operational challenge of managing and deploying large language models at scale has given rise to a specialized field called LLMOps, an extension of MLOps that addresses the unique complexities of LLM lifecycles, from versioning and monitoring to managing costs and governance. - For future releases like iOS 27, Apple is working on a more advanced project codenamed "Campo" to create a true chatbot-style assistant, which is expected to rely more heavily on Google's servers and a more advanced custom Gemini model. - The push for advanced AI assistants is mirrored in the insurance industry, where large language models are increasingly being applied to actuarial work for tasks like analyzing unstructured claims data and enhancing risk modeling. - The growth in AI is fueling the tech scene in New York, which is now home to over 2,000 AI startups and more than 40,000 AI professionals, supported by initiatives like the $400 million public-private Empire AI consortium.