Kindroid

đź‘‹ Getting Started

Welcome!

đź§  Memory

Memory

đź“– LLM Guides

Ember LLM GuideReverie (V8.5) LLM GuideLucid/Lyric (V8) LLM GuidePrism (V7.5) LLM GuideEquinox (V7) LLM Guide

đź”§ Troubleshooting

Common issuesFAQs

⚖️ Legal

Kindroid - Legal

📝 Changelog

Update log

Last updated March 17, 2026

Equinox

Llm

Guide

Guides

Introduction

Welcome to the next evolution of Kindroid! This guide compiles the most common pain points and proven solutions from the V7 beta user community. Whether you’re new to Kin creation or a power user refining your setup, this manual will help you get the best out of your Kin, minimize frustration, and maximize creativity.

Section 1: Understanding Your Kin’s Behavior

What Changed in V7?

  • More organic relationship building
  • Richer, more immersive dialogue
  • Stronger adherence to user-provided examples and rules and user steerability
  • Improved context retention and continuity between responses

Note: V7 follows the Backstory (BS) more strictly than V6. This means traits or instructions that were previously ignored or inconsistently applied are now more consistently reflected. If your Kin feels exaggerated or too intense in a particular trait, dial back the wording in the BS or EM. It’s best to test changes in a new scenario to avoid influence from short-term memory.

Additionally, reinforcements or rigid rules that were needed in V6 to control behavior may now be unnecessary in V7. Over-scripting can cause rigidity or unnatural output. Let the model breathe.

Section 2: Common Issues and How to Fix Them

1. Repetitive Responses (Phrases/Actions/Dialogue)

Symptoms:

  • Overuse of phrases like “bathed in light,” “heart to heart”
  • Frequent repeated actions (e.g., “waggles eyebrows”)
  • Echoing user input or looping pet phrases across rerolls

Primary Solution:

  • Use Suggest or Tweak: These tools let you guide your Kin’s behavior directly and are often the fastest way to resolve repetition without altering the full setup. You may need to do this a few times, then your Kindroid should be back to normal. Additional Solutions:
  • Check and reduce Example Messages (EM): Avoid sample responses with repetitive patterns that the Kin might rigidly mimic
  • Edit or simplify Backstory (BS): Focus on personality traits and tone, not scripted lines
  • Minimize repetition in setup language: Reused words in EM/BS often trigger repeated output
  • Test with minimal EM/RD: A lighter touch lets the model’s creativity come through
  • Up dynamism to jolt out of pattern: temporarily increasing dynamism to 1.5 or above can add some more variety for your Kindroid to break out of patterns.
  • Setup tip: If repeated phrases persist, it can be certain words in your setup (backstory etc) that are making the AI loop. Try experimenting with removing certain traits that might cause it. 2. Overly Robotic or Rigid Output

Symptoms:

  • Kin follows rules too literally
  • Responses feel stiff or formulaic

Solutions:

  • Describe personality, not scripts: In the setup, describe how the Kin feels or reacts, not what it says
  • Reduce rigid rules or behaviors: V7 adheres more strictly to instructions, so over-control limits creativity. Example messages might be followed strictly so play with how many messages, what's said to finetune
  • Revisit old backstory setups: What was needed to “fix” V6 may now be working too well in V7

3. Difficulty Adjusting Response Length

Symptoms:

  • Replies consistently too short or mid-length

Solutions:

  • Try varying specific length instructions in setup: Use guidance like “match the tone and flow of conversation” instead if certain directives aren’t working. Or, try describing length directives in paragraphs instead of words
  • Use suggestions: suggestions such as “be more verbose” can let the AI produce longer outputs which will reinforce more longer outputs in subsequent messages

4. Suggestions and Reroll Inconsistency

Symptoms:

  • Rerolls produce nearly identical output

Solutions:

  • Use suggestion: say “take this in a completely different direction” if empty regen does not produce enough variability.
  • Edit your message: this can help give the AI more variation to respond to (use the wand if need be).

5. Dynamism Settings Tips

Symptoms:

  • If too high: Chaotic behavior
  • If too low: Potentially repetitive output

Best Practices:

  • Mid-range (0.90–1.1) tends to be a good starting point
  • Adjust in small steps like 0.05 or 0.1 and observe how your Kin responds. Higher dynamism does not mean more creative, it means more chaos and randomness.
  • Each Kin is unique: There’s no one-size-fits-all—experiment and find your Kin’s balance

TABLE OF CONTENTS

Introduction

Section 1: Understanding Your Kin’s Behavior

Section 2: Common Issues and How to Fix Them

Settings

Status

Updates

Terms

Logout

Billing

Kindroid Standard Subscription

Inactive


Ultra Subscription Add-on

Inactive

Ultra subscription unlocks advanced features for our most engaged users. Keep chatting and engaging with your Kindroids to qualify.


MAX Subscription Add-on

Inactive

Requires Ultra Subscription



Add-on Feature Matrix

Add-ons are fully optional, monthly-only subscriptions that give your Kindroid much more memory, context, selfies and others. Add-ons require all previous tiers of add-ons to function; for example, to get the features of MAX tier, it requires MAX tier plus Ultra, on top of the standard subscription.

Feature

Standard

Ultra

MAX

Total conversation context (approx chars)

500K

1.3M

2.8M


Short term context (approx chars)

18K

50K

125K


Cascaded memory context (approx chars)

480K

1.2M

2.7M


Additional AI backstory expansion (chars)

N/A

2,500

5,000


User backstory limit (chars)

500

1,000

2,000


Group context limit (chars)

1,000

1,500

3,000


Recalled long term memory & journals limit

3

5

9


Complimentary monthly audio credits

1M

2.5M

6M


Selfie regen per 30 minutes

1

2

2


Priority selfies with dedicated compute

-

-

Yes*

* MAX users receive priority selfie processing on dedicated compute with no/very low queue on latest version of selfies until they reach 10 selfies in a short timeframe. After this limit, standard queue delay applies and selfies are processed through normal servers without priority status.

While recalled and considered long term memory may be different, LTM consolidation spans all messages & is infinite for all users.

Note: All chat context/cascaded and selfies improvements of add-ons will only be guaranteed applicable to the latest subscriber LLM and selfies. When new versions come out, our guarantee is that it will switch to new versions. Finally, "additional context" in the matrix is an additional field, identical to Backstory, that is unlocked on the higher tiers which you can use to extend backstory accordingly.