Hi Stephen,
thank you for taking the time to share your feedback.
We’re genuinely sorry to hear about your disappointment — every piece of feedback matters to us, and listening to our users is a core part of how AIRO improves.
That said, we’d like to clarify a few important points, as timing plays a role here.
When GPT-5 was first released, we intentionally did not migrate immediately. In our real-world testing at that time, early versions of GPT-5 proved unstable for professional workflows (including critical issues such as unreliable document uploads and inconsistent outputs). For an expert-layer product like AIRO, this was not an acceptable trade-off.
With the recent release of GPT-5.2, this changed.
Just in the last few days, we have:
• Refined the internal logic and prompts of every assistant
• Updated the recommended models per assistant, based on their role (GPT-5.2 Thinking / GPT-5.2 Instant / GPT-5.2 Standard)
• Selected each model strategically rather than applying a single model across the board, to ensure the best possible performance per use case
AIRO is not designed to compete with ChatGPT as a general-purpose chat. It works as an expert layer — intentionally more structured, more direct, and sometimes more challenging — closer to how a senior professional would reason and guide decisions.
If your test happened before or during this transition, we sincerely invite you to try the assistants again now, using the newly updated model recommendations. Many users are already seeing a clear improvement in depth, clarity, and reliability with GPT-5.2 correctly applied.
That said, we fully respect that different users prefer different working styles. Your feedback helps us improve both the product and how we communicate these changes.
The AIRO Φ Staff
