Apple Builds Internal ChatGPT-Like Tool to Test Siri Features

Apple is quietly working on something big behind the scenes. According to a Bloomberg report by tech reporter Mark Gurman, the company has created an internal app - code-named Veritas that works like ChatGPT. This app is used by Apple’s developers to test new features for Siri before these changes are introduced to regular users.

What is Veritas, and what can it do?

Veritas lets Apple engineers try out advanced features so they can see what works or doesn’t. It reportedly supports multi-turn conversations, meaning you can ask follow-up questions, and Veritas remembers the context, similar to ChatGPT.

Some tasks being tested include letting Siri search your personal data, like emails or songs, when you ask it to find something or doing things inside apps, for example, editing photos on command. Right now, Veritas is internal only. 

Siri’s Overhaul Delayed Until 2026

Apple’s overhaul of Siri has been in the works for a while. The plan was to make Siri smarter, more conversational, able to act across apps, and better at understanding what you mean. However, this upgrade has been delayed and may not arrive until spring 2026.

Sources say Apple’s internal testing revealed performance issues. Some executives reportedly worried the new features weren’t working as well as expected, forcing Apple to slow down and rework aspects of Siri’s redesign.  One outlet even suggests part of the redesign may have to be rebuilt from scratch.

Given the delay, Apple’s not rushing to release anything before it’s ready, likely to avoid a poor user experience or privacy missteps.

What’s Next And Why This Matters

Veritas is effectively a proving ground. By trying new ideas internally, Apple can catch problems early and avoid pushing updates that are buggy or unsafe.

If the revamped Siri eventually launches in 2026 with features tested via Veritas, users might see a much smarter assistant, one that remembers past questions, understands context, and acts across apps. But if internal testing doesn’t hold up, parts of the project might change or be removed altogether.

Read More: OpenAI Pilots ChatGPT Pulse - A New Feature For Personalised Updates