This update brings some big improvements to how your AI characters understand what's happening physically, including full position awareness, smarter touch descriptions, and proper support for scenes with multiple characters.
Position Tracking
Your AI characters now know what position they're in.
The plugin detects eight single-character poses - standing, kneeling, sitting, bent over, on all fours, lying on back, lying face down, and lying on side - and seven two-character positions during intimacy (missionary, doggy, cowgirl, spooning, standing, bent over, and kneeling oral).
When your character changes position, the AI is informed. Outside of intimacy this comes through as silent context so the AI is aware without interrupting. During intimacy position changes trigger a reaction so your character responds when you move them into a new position.
Composite positions like "missionary" or "doggy" only get named when penetration is actually happening — so just having your character on all fours won't trigger assumptions about what's going on.
Four settings let you tune it to taste:
- Position Tracking on/off
- Position Cooldown — how often position updates can fire (default 5 seconds)
- Position Confirm Time — how long a pose needs to hold before it counts (default 0.7 seconds, filters out flickering during transitions)
- Position Fire Once — recommended for most users. Sends exactly one message per position change, no repeats
Smarter Touch Descriptions
The AI now knows what body part is doing the touching, not just what's being touched.
Before, a touch on the breast could mean anything — hand, mouth, hip, whatever. Now the message tells the AI "with their hand" or "with their mouth" so it can react appropriately. A hand on the breast is different from a kiss, and the AI finally understands that.
Body parts described: hand, mouth, head, chest, foot, knee, thigh, penis. More ambiguous contacts (like someone's hip brushing another character) just report the zone being touched without speculating.
This works whether you're in VR, desktop, or embodying a Person atom directly.
AI-to-AI (Person Atoms) Support
You can now run scenes with multiple AI characters interacting with or without you.
With the new
"Attribute touches to source character" toggle on the Touch tab, when Person1 touches Person2, the AI is told "Person1 touches Person2" instead of the default "{{ user }} touches {{ char }}". The plugin looks up which Voxta character each atom corresponds to and fills in the right names.
This works for touches and for all three orifices (vaginal, anal, oral) - entry, ongoing descriptions, and exit messages all carry the correct character names.
Plus, a big fix for multi-character scenes: the plugin now correctly identifies which character is being touched or penetrated. Before, if you had the plugin on two different characters, messages could get mixed up and both report as happening to the same character. This has been fixed - each character's plugin instance properly tracks its own identity.
Better Detection
Position detection reads directly from the character skeleton rather than from control points. This means it works accurately whether you're manually posing in VR, playing animations, using MoCap, or running scene logic that drives the character through physics.
Improvements to Existing Features
- Depth tracking got faster intensity detection. Thrusting speed is now measured properly, so Hard and Deep intensity messages actually fire during vigorous activity instead of being stuck on Soft
- Longer cooldown options. All message timing sliders (touch, body part, depth, position) now go up to 60 seconds for users who want less chatty AI reactions
Compatibility
All new features default to sensible settings that match the old behaviour. If you don't touch anything, scenes work as before. Enable the new features as you want them.
If you use ATA (Attraction, Touch, Arousal), an update version can be found on the Voxta Discord.
