Natural language processing lagging behind OS native capabilities
Constructive feedback time! I've been a Fantastical user for years and really liked the NLP processing feature when I first use it.
But it feels frozen in time, and I think that the team should upgrade it. I can't help but feel like the "NLP" is still rules-based rather than "smart" in the "local natural language model running on my iPhone" sense users expect in 2026.
It doesn't do a good job of detecting things like "Client Acceptance Test - Meeting 2/4 tomorrow" contains 2/4 as a part of the title, rather than a date of Feb 4, and the meeting is actually for tomorrow.
It also doesn't run through a confirmation step to check for errors like the chat based tools do, which has led to multiple "vanishing" calendar events for me (for example: I create an event for yesterday's date that actually gets created for the same day next year). This, combined with the lack of any obvious way to view recently created events, makes it hard to debug what went wrong when an event "vanished".
I understand Devs may want debug information in when the behavior is occurring, but the problem with requesting that from your users in 2026 is that it shouldn't be necessary. The expectation from users is that software is less deterministic, and users are used to the idea that it will work more often than not, especially for apps advertising AI capabilities. Rules-based golden paths are no longer a user expectation, especially for fairly expensive subscription software.
I still like Fantastical a lot, and prefer its interface over competitors like Apple Calendar. I'd like to see it improve and the Flexibits team continue to succeed :)