Introduction
For years, Apple has avoided chasing every new trend, often arriving late but reshaping categories once it commits. Now, as Google’s Gemini and OpenAI’s ChatGPT redefine how people use technology, Apple is preparing its own system to keep pace. Reports suggest Apple has been testing a ChatGPT-style prototype internally (codenamed Veritas), while refining what it calls “Apple Intelligence,” a suite of system-level features built on new foundation models. By blending on-device and cloud models, Apple is trying to enhance Siri and everyday iPhone experiences.
What Apple Is Building
Apple’s work revolves around two key layers:
- On-device models: 3 billion parameter on-device language model which is efficiently designed to run directly on iPhones, iPads, and Macs. They can summarize emails, rewrite messages, or generate fun images for chats without ever sending personal data outside the device.
- Server-based models: Heavier, more capable models that live on Apple’s servers. They handle tasks requiring more computing power, such as advanced writing assistance or pulling broader knowledge from the web. These are tied together with a private compute layer designed to minimize exposure of personal information (Source).
Internal Testing: Project Veritas
As per Mark Gurman (Managing Editor at Bloomberg), Apple is testing Siri’s upcoming revamp using an internal chatbot called Veritas. Think of it as Apple’s in-house version of ChatGPT but not built for public release (yet). Apple’s employees use it to test conversation flows, query their own device data, and evaluate how well the models handle multi-turn interactions. This new Apple’s “Veritas” app can shape the future of Siri’s capabilities.
Why Apple Is Doing This
There are three key motivations behind Apple’s efforts:
- Changing user expectations — people now expect assistants that can hold conversations, not just handle one-shot commands.
- System integration — Apple prefers embedding features into its ecosystem rather than releasing standalone products.
- Privacy as a differentiator — Apple emphasizes on-device models and limited server use.
How Apple’s Strategy Differs from Rivals
- Hybrid design: Google and OpenAI lean on the cloud; Apple blends local and server approaches.
- Embedded features vs standalone apps: Apple enhances iOS/macOS instead of launching a new chatbot.
- User’s Privacy protection: Since it uses local on-device model, and for further processing it does not rely in any open server, instead using their own private cloud compute. Thus, it will build user trust.
The Challenges Ahead
Apple faces hurdles despite its vast resources:
- Catching up in quality: Gemini and ChatGPT already have public momentum, Apple may face hurdles in catching up with the quality of the model they came up with. Since Google and Open AI have the early mover advantage, and many people are already using their models. So, their models have trained to a great degree, whereas Apple’s model will be new.
- Talent wars: Apple must recruit and retain top researchers, as Apple AI researchers were moved to OpenAI, Meta and Anthropic in the past.
What to Expect Next
With the introduction of this new Apple’s AI-enabled Conversational model, the following improvements are anticipated in Apple’s products: –
- Expect smarter Mail, Messages, and Siri improvements.
- Siri may soon provide richer, world-knowledge answers.
Conclusion
Apple’s entry into the conversational space is less about launching a rival to ChatGPT, and more about making devices feel more helpful, private, and natural to use. That’s Apple’s classic playbook: avoid hype, refine experiences, and let the product speak for itself.
