The age of AI seems to forget some lessons from Google (and history in general).

- Rapid changes hurts the trust of your brand and product. In Google case, using a new service product became something you’ll think multiple times as you are more likely to axe it than rivals or specialized equivalents.

- While models currently has no clear winner. Anthropic’s core product is coding. But just as Skype, IE, Netscape their can always be another game changer you cannot count.

- The Pro plan is already limited for true agentinc workflows. The limits now are so bad that a business that relies on it would need bigger plans.

- Anthropic is already in a delicate situation where many devs are frustrated. Dropping or crippling the use even more just means this sector (which I can only assume is a big chunk) would switch to competitors tool that already try to compete.

- Local models, whether as Google sees it “edge” or even further would also take bigger part in the future.