Odyssey AI
Odyssey AI
The journey to hyper-personalisation. How Odyssey primes large language models with structured context to create emotionally intelligent financial experiences.
Where The Journey Began
Artificial Intelligence presents enormous opportunity to explore the world, identify patterns within it, and deploy intelligent systems to do work that is boring, complex, or dangerous. AI works by harvesting large amounts of data, evaluating patterns that can be described mathematically, and pointing those patterns in the direction of good. At Moroku, we believe that good involves using AI to help people thrive with their money.
Moroku CEO Colin Weir learnt the foundational importance of data during his first attempt at building algorithms in the 1980s. His post-graduate research focused on developing mathematical models to determine the economic impact of management decisions on the plantation forests of New Zealand. Past knowledge suggested that yield curves had a sigmoidal shape, describable by a yield equation with various coefficients representing the impact of yield (Y) over time (T).
As more data was collected and coefficients analysed, a new parameter emerged: competitive index. This single addition dramatically improved prediction accuracy. The key insight was clear; algorithms are only as good as the data and parameters that feed them.
During the 1980s, this research was conducted manually, laboriously adding and testing parameters. The power of modern AI lies in its ability to continue adding and testing parameters automatically, learning the connections between inputs and outputs at unprecedented speed. By continuously refining these relationships, AI creates algorithms that self-adapt to fit data distributions—the foundation of machine learning.
Data Sets
The foundational datasets upon which the model is trained and learns patterns
Algorithms
The depth and learning ability of the mathematical models to identify patterns
Purpose
The direction and bias with which the AI is oriented toward outcomes
Priming LLMs for Financial Wellness
Odyssey creates the initial algorithm sets that prime large language models around a telos of engaging, personalised financial wellness experiences
Data Integration & Aggregation
Odyssey collects and integrates financial data from multiple sources, creating a comprehensive view of each customer's financial health. Transaction data flows from core banking systems, while Open Banking connections provide visibility across the customer's complete financial landscape.
- Real-time transaction event streaming
- Open Banking aggregation via Biza.io
- Behavioural pattern recognition
- Psychological archetype profiling
Player Map Instantiation
The player map provides a multi-dimensional coordinate system that positions each customer within a structured framework. This isn't crude demographic segmentation, it's precise positioning across leagues (earning, spending, saving, lending, investing, helping), missions, levels, and psychological archetypes.
- 16,807+ unique coordinate positions
- 7 financial leagues with nested missions
- Life stage and goal tracking
- Continuous position updates as customers progress
LLM Context Priming
Rather than letting LLMs generate generic responses, Odyssey provides structured context that constrains and directs output. The player's current position, recent trajectory, psychological profile, and available actions all feed into the prompt, ensuring relevance and appropriateness.
- Defined state space for reasoning
- Constrained action space via nudge library
- Clear reward signals for optimisation
- Compliance guardrails embedded
Continuous Learning & Refinement
Every customer interaction provides feedback that refines the algorithms. The system learns what motivates each individual, which nudges drive action, and what reward structures resonate. Over time, generic cohort-based rules evolve into genuinely individualised algorithms.
- Response tracking and outcome measurement
- A/B testing of nudge variations
- Financial fitness score evolution
- Per-customer algorithm refinement
From Cohorts to Individuals
Odyssey begins with cohort-based algorithms and evolves toward genuine hyper-personalisation as the system learns each customer
Initial Algorithms
Cohort-based rules spread across representative player leagues, placing customers on the map based on demographics and relationship history
Data Enrichment
Psychological archetyping and Open Banking data grow the dataset, increasing algorithm granularity and pattern exploration
Individual AI
Deep learning takes over algorithm generation as each customer develops algorithms refined specifically for their context
Beyond Logic and Reason
At the heart of this journey toward hyper-personalised experiences lies a requirement that transcends data and algorithms: the capacity for empathy. True engagement doesn't come from optimising logic or perfecting reason. It comes from connecting to people emotionally, from understanding that we act not based on what we think, but on how we feel.
"We know we act based not on what we think but on how we feel. To improve its utility, AI must go beyond trawling for previously defined answers to understanding the human factor."
Much of human behaviour comes down to one fundamental drive: making ourselves feel better. Sometimes this aligns with genuine improvement; sometimes it doesn't. We buy things we don't need because the purchase makes us feel good. We avoid checking our bank balance because ignorance feels better than confronting reality. We celebrate small wins out of proportion to their significance because celebration feels rewarding. Understanding this truly understanding it is the difference between systems that inform and systems that engage.
Emotions are the foundation of relationships. Not transactions, not interactions, not touchpoints, but relationships. When a customer feels understood, they stay. When they feel celebrated for their progress, they engage more deeply. When they feel supported through difficulty rather than lectured about failure, they trust. These emotional connections cannot be manufactured through clever copy or optimised through A/B testing alone. They emerge from systems designed with emotional intelligence at their core.
This remains AI's greatest challenge and its greatest opportunity. For those looking to use ChatGPT to complete exams, apply for jobs, or respond to RFPs, there remains a long way to go. For those seeking to create genuine connection. The kind that builds loyalty, drives engagement, and transforms customer relationships. The path requires something more than pattern recognition. It requires understanding that the goal isn't to be right; it's to make people feel understood, supported, and capable.
AI is amazing. Beyond stripping cost out of customer service and generating content at scale, there is real potential to help customers thrive in the new digital world. But that potential is only realised when we remember that thriving isn't a logical state. It's an emotional one. Odyssey exists to bridge that gap: providing AI with the structure to be accurate and the context to be emotionally intelligent, creating experiences where customers don't just succeed with their money, but feel good about the journey.
Celebration
Recognising achievements creates positive emotional associations with financial progress
Support
Being present during difficulty builds trust and deepens the relationship
Empowerment
Helping customers feel capable drives engagement more than instruction ever could
Harness AI for Customer Success
Place your data onto the Odyssey player map to take your customers on their money journey, with AI that understands both the numbers and the emotions.
