AnkiGenix v1.3 Update

v1.3 is a release centered on two things: generation quality and production reliability.
Instead of shipping a large number of surface-level features, this update focuses on a few changes that have a direct impact on real usage:

• card planning for long and structured content needed to be more accurate
• Auto mode still had edge cases in language-following behavior
• signup onboarding for new users was still incomplete
• APKG export could fail in some production deployment environments

If you regularly generate cards from long articles, terminology-heavy material, multilingual content, or depend on export as part of a stable study workflow, the improvements in v1.3 should be noticeable.

Highlights

1. Smarter chunk-level card planning

v1.3 upgrades how longer content is analyzed before cards are generated.

The system now does a better job estimating:

• whether a chunk is mainly structured content
• how dense the knowledge is inside that chunk
• how many cards that chunk should realistically produce

This is not just an internal optimization.
It improves the actual output quality, especially for lists, glossary-like sections, Q&A-style material, and longer chapters.

In practice, that means:

• fewer missed points
• fewer duplicate cards
• fewer cards that try to test too much at once

2. More reliable Auto language behavior

We continued refining the language flow in Auto mode.

This release makes a clearer distinction between ordinary natural-language wording and technical terms that should remain in English.
The result is that card wording now follows the dominant source language more consistently, instead of drifting toward English too easily in Chinese, Japanese, or Korean content.

Exact product names, commands, APIs, protocols, and canonical technical terms can still stay in English when needed.
But the surrounding question wording, explanations, and connectors now stay much closer to the original language.

That makes multilingual review feel more natural and more trustworthy.

3. Welcome email for new users

v1.3 also improves the signup experience.

New users now receive a Founder Welcome Email after registration.
The goal is simple: make the first-touch experience clearer, more guided, and more complete.

We also changed the flow so registration bonus credits and welcome email sending run in parallel.
If email delivery fails, it does not block the signup flow itself.

4. More stable APKG export in production

We fixed a lower-level but high-impact issue in the export path:
in some production environments, the sql.js wasm asset used for APKG export could fail to load correctly.

When that happened, APKG export failed outright.

v1.3 adds a more reliable runtime lookup path for that dependency and adjusts the production startup flow accordingly.
If you depend on APKG export to move content back into Anki, this update should make that workflow much more dependable.

What this means for you

If you mostly generate cards from short and simple input, v1.3 will mainly feel more stable.
The gains are much more visible if your workflow includes:

• long documents or chapter-based material
• list-heavy or terminology-dense content
• Chinese, Japanese, Korean, or other non-English source material
• APKG export as part of a serious review pipeline

What’s next

After v1.3, we will keep pushing in the same direction:

1. improve card atomicity and coverage even further
2. keep reducing duplicate and overly broad cards
3. keep improving multilingual and technical-content generation
4. continue polishing export reliability, task feedback, and overall stability

If you are already using AnkiGenix, keep sending us real outputs from your workflow.
That kind of feedback is still the fastest way for us to decide what to improve next.