No response needed. Just providing info of things that have happened since I first joined. These things are not just specific to this one adventure.

Memory Degradation & State Loss

Memory reliability is also a major problem.

It appears that no matter how many instruction blocks, memory notes, or context reminders are provided, Franz’s memory degrades rapidly within a single session.

I have had to remind Franz over a hundred times during one three hour session about:

• party location

• current island

• recent fights

• completed objectives

• defeated enemies

• ongoing storyline arcs

Despite these reminders, Franz repeatedly forgets established state and attempts to:

• move the party back to locations already completed

• reintroduce resolved conflicts

• reset progress

• contradict previously established events

This results in constant continuity breaks and forces the player to repeatedly correct the AI.

For a system that is supposed to function as a Game Master and track world state, this level of memory instability makes longer sessions impractical and undermines confidence in the platform.

===================================

Canon Accuracy Claims vs. Actual Capability (Contradictory Behavior)

There is also a significant reliability and trust issue regarding Franz’s stated capabilities versus its actual behavior.

I fully understand that Friends & Fables is designed primarily for custom, player-created adventures and is not inherently required to replicate any existing storyline exactly. I recognize that the platform’s intended use is to support original campaigns rather than strict recreations of established media.

However, when I explicitly requested that my campaign be treated as a faithful, canon-accurate experience based on the One Piece storyline, Franz repeatedly confirmed that it would run the adventure aligned with the official arcs, events, and characters and that it would handle those canonical story beats directly without requiring me to manually supply them.

These responses clearly implied full or near-complete canon knowledge and the ability to execute the storyline accurately.

Later, when questioned directly about its knowledge scope, Franz admitted that it does not have complete canon knowledge and instead operates on approximations.

This creates a direct contradiction:

• Franz claims it can run strictly canon-aligned storylines

• Franz later admits it relies on approximations and incomplete knowledge

Because the system lacks definitive canon awareness, it necessarily fills gaps with guessed or fabricated details, which leads to:

• incorrect sequencing

• invented or altered events

• inconsistencies with established lore

• deviations from expected story progression

From a technical standpoint, strict canon fidelity is therefore not achievable under these constraints.

The issue is not that Franz cannot perfectly replicate canon — that limitation is understandable. The issue is that Franz explicitly stated that it could do so and configured the campaign as if this were guaranteed. This mismatch between stated capability and actual behavior is misleading and undermines user trust.

If the AI cannot reliably deliver canon accuracy, it should communicate that limitation clearly and consistently rather than implying full fidelity and later retracting that capability.