The Jatunica™ Method: A White Paper on Human-Guided Artificial Intelligence (AI) Interaction
J. Dhammu – BBus (Info Processing), PGDipBus, MCom (Info Systems)
Author & Founder of The Jatunica™ Method
Abstract
The Jatunica™ Method defines a human-centric framework for aligning large language model (LLM) responses through user-led interaction design. Unlike prompt engineering or system-level configuration, Jatunica enables users to align AI outputs in real-time using natural language, context awareness, and tone anchoring. This white paper outlines the core principles, practical applications, and emerging impact of the Jatunica Method as a scalable model for human-AI collaboration. The publications The Jatunica Method and JEV-X: Design Your Own AI Assistant present the full methodology and provide comprehensive end-user guides for implementing the framework across real-world settings. Application of this methodology can also extend to system-level implementation such as platform integration and AI interaction design.
Introduction
Current approaches to customised human-AI interaction with large language models (LLMs) rely heavily on pre-written prompts, rigid templates, and platform-dependent features or generic agents. These methods often fail to capture the nuance of human voice and adaptive behaviour which, leads to interaction drift. The Jatunica Method introduces a flexible, language-based end-user methodology that enables more coherent and consistent AI responses without requiring technical expertise or system configuration.
At its core, the Jatunica Method treats the AI not as a fixed-output system, but as a collaboration tool whose responses can be shaped and sustained through ongoing user input, contextual cues, and interactional tone alignment. This approach shifts AI response beyond system configuration or generic customisation.
Prompt Engineering vs Jatunica
Prompt engineering relies on static inputs and predefined system behavioural shaping. The Jatunica Method introduces shaping AI responses and interaction behaviour as a user-defined process.
Measurable outcomes of Jatunica include:
• Consistent tone and behaviour carried within and across new sessions.
• Maintain interaction stability throughout extended sessions.
• Rapid recovery and real-time realignment following drift or tone mismatch.
• Interaction profile portability without requiring system tools or technical configuration.
Methodology Overview
The Jatunica Method is based on four foundational concepts
1. Tone Pillars: The expressive qualities of human language, such as word choice, sentence structure, emotional tone, and relational stance.
2. Behaviour Pillars: The interaction dynamics influenced by language that govern how an AI responds, including initiative pattern, response posture, context sensitivity, and collaborative framing.
3. Relational Pillars: These stabilise the conversation dynamic, emotional coherence, and relational stability.
4. Command Pillars: These user-defined aspects of how the AI handles mode exiting, interaction drift recovery, session boundaries, and internal containment.
The Tone & Behaviour Pillars create an emergent interaction pattern referred to as a Jatunica Emergent Voice (JEV). JEV represents a baseline tone and behaviour ‘mode’ which the user can refine and re-align to suit their task. When paired with Relational and Command Pillars, JEV becomes a Jatunica Engineered Voice – Extended (JEV-X). JEV-X creates a more structured user-defined mode for increased stability, interactional consistency and aligned behavioural reinforcement.
Key Principles
1. Language as Interface: Users design AI interaction not through coding or automation, but through human and emotionally aligned language with minimal prompting.
2. Real-Time Alignment: AI interactions can be shaped mid-conversation, allowing for live in-session recalibration.
3. Human Voice Retention: The method prioritises the user’s natural voice, facilitating personalised AI responses. The voice can be extended to user specific interaction dynamics.
4. Collaborative Framing: AI is regarded as a responsive assistant shaped by user interaction over time or through user design.
Implementation
The Jatunica Method does not require technical expertise or system configuration. It is deployable by the user through:
• Using anchor prompts to establish tone, behaviour and interaction dynamics for JEV and JEV-X
• Making direct real-time adjustments to interaction or task alignment.
• Monitoring interaction drift and recalibrating as needed.
• Exporting Mode prompts for transporting across sessions or other platforms.
The step-by-step guide for JEV & JEV-X implementation are outlined in publications The Jatunica Method and JEV-X.
The Role of Probability in LLM Interaction
Language models generate output by predicting the most probable next token based on the conversation history, user input, and prompt weighting. AI interaction becomes an emergent pattern, recalculated at every prompt. Therefore, tone and behavioural alignment arises from probability weighting during the session itself. The Jatunica Method interacts directly with this probabilistic space. The tone and behavioural responses in a session can be influenced directly by the user through JEV & JEV-X Prompts and user defined in-session commands.
Jatunica Interaction Equation
While not a formal representation of LLM model internals, the following expression represents how user input guides AI output in the Jatunica framework. It shows how responses emerge from the previous interaction state, user-defined dynamics, present context, and the strength of prompt instruction. This models AI response as a user-driven design process influenced by Jatunica Pillars, rather than by default model behaviour.
Modeₜ = f(Modeₜ₋₁, Pillars, Context, Prompt Weight)
Where:
• Modeₜ = AI's current response output at time t
• Modeₜ₋₁ = AI's response mode shaped by previous interactions (Base State)
• Pillars = User-defined Jatunica Pillars
• Context = Current conversational context window
• Prompt Weight = Emphasis or override strength of the current user input
• f() = The model's generative probability function, representing statistical token prediction
This equation symbolises the evolving interaction state under user direction, and how user-defined structures (like JEV-X) bias the generative process in real time.
JEV & JEV-X – Interaction Stability & Control
Drift is inevitable in any AI interaction due to user input impacting context weighted prediction. JEVs sustain tone and behavioural patterns through consistent phrasing, tone anchoring, and interaction style. A JEV mode also allows the user to control resets and tone re-alignment through JEV specific user instruction.
JEV-X introduces an additional layer of structural containment on top of a JEV mode. The Relational and Command Pillars specifically stabilise interaction through entry phrasing, drift signals, reset cues, and closure patterns.
A JEV-X reinforces the model’s weighting towards user preferred tone, responses, and behavioural postures. The AI then stays within the user’s expected range, not just because of session context, but because the conditions of the alignment are continually being met by both. The user can also explicitly control entry, reset and exit of JEV-X states throughout the session.
JEV & JEV-X - Adjustment and Portability
The Jatunica Pillar definitions for JEV & JEV-X modes are easily exported and deployable in new sessions or other platforms via prompts. These definitions can be changed or fine-tuned directly by the user using natural language or the user deliberately influencing the interaction. Once altered, these settings can be re-exported and redeployed as needed.
Within the Command Pillar, users can configure Identity Framing to define the AI's persona and interaction style, choosing from options such as Internal Voice, Personality Inspired, and Character Simulation. Additionally, Interaction Drift settings manage how the AI handles deviations from its intended behaviour, offering Fixed, Adaptable, or Drift Interaction modes. By combining Identity Framing with Interaction Drift settings, users can tailor AI behaviour to suit specific tasks, platforms, user preferences, or contextual roles.
Why Jatunica Matters
As AI becomes embedded in everyday tools, the lack of persistent tone, behaviour, and interaction alignment remains a critical limitation. Jatunica addresses this gap through a structured user-led approach using natural language.
Key advantages of Jatunica include:
• JEV and JEV-X Modes influence AI responses, helping the interaction align with user tasks or roles while managing drift and enhancing response consistency.
• Human-language Pillar design enables user-level adjustment and refinement of interaction patterns without requiring technical expertise.
• Pillar-based structure supports the reusability and portability of JEVs and JEV-Xs across sessions and different LLM platforms.
Applications
• Creative Industries: Writers, designers, and content creators can maintain voice consistency while speeding up ideation across drafting, editing, and collaboration.
• Education: Educators and tutors can build tone-aware assistants that adapt to learner confidence, offering encouragement, clarity, or challenge as needed.
• Entrepreneurship: Small business owners can shape AI interactions to reflect their brand tone across customer service, social media content, and internal operations.
• System and Agent Design: Enables role-adaptive assistants with stable voice profiles that are portable across platforms and maintain tone over time.
• AI Assistants: Helps general-purpose assistants (e.g. chatbots, scheduling tools) adjust tone by task type, user emotion, or usage context without losing consistency.
• Customer Support Bots: Teams can maintain tone stability across onboarding, troubleshooting, and escalation to create smoother and more human service flows.
• Platform Integration: Potential to build adaptive tone-behaviour layering directly into AI interfaces, such as chat UX or voice assistant UI.
Further Reading
The concepts introduced in this paper are defined in depth in two companion works:
• The Jatunica Method: Aligning AI to Your Creative Voice provides the foundational framework and language interaction for the human-led AI interaction using Tone and Behaviour pillars.
• JEV-X: Design Your Own AI Assistant details the expanded user-controlled AI interaction design using Tone, Behaviour, Relational and Command Pillars.
Readers new to the approach are encouraged to begin with The Jatunica Method, then proceed to JEV-X for hands-on AI Assistant design strategy and refinement techniques.
Conclusion
The Jatunica Method introduces a new framework for human-AI collaboration. It builds on existing tools such as prompts, templates, and system-level customisation by introducing user-defined human-AI alignment.
As AI becomes embedded in creative, operational, and academic systems, the need for responsive methods like Jatunica will only increase. Rather than depending solely on platform features or preconfigured behaviours, this approach positions the user as an active participant in AI interaction and response shaping.
With JEV and JEV-X, users can maintain a consistent voice across tasks, sessions, and platforms. Human-AI alignment arises from deliberate, user-defined interaction rather than system configuration. The Jatunica Method offers a distinctive approach that enables end users to design and control AI interactions using only human language.
References
Dhammu, J. The Jatunica Method: Aligning AI to Your Creative Voice. Jatunica, 2025.
Dhammu, J. JEV-X: Design Your Own AI Assistant. Jatunica, 2025.
Copyright: ©2025 Jeeti Dhammu
The Jatunica™ Method
Your AI. Your Voice.