Inspiration
Our inspiration stemmed directly from NatWest's call to champion potential and build financial resilience for underserved communities. We realized that traditional budgeting apps often fail those who need them most—such as time-poor carers, neurodivergent individuals, or those from lower socioeconomic backgrounds—because they rely on passive tracking and induce anxiety. We wanted to eliminate the cognitive load and shame associated with money management by creating an empathetic, voice-first AI that feels like a supportive partner rather than a rigid, judgmental ledger.
What it does
NatAware is a proactive financial resilience platform centered around a conversational AI named "NatBot." Users interact via a highly accessible voice interface, while a hidden backend algorithm assesses their psychological spending archetype without asking intrusive questions. Instead of just showing past transactions, the dashboard cross-references the user's data with predictive, real-world variables—such as localized weather changes or seasonal health trends—to offer proactive nudges, automatically recommending micro-transfers into designated "Vaults" (like Essentials, Splurge, and Safety Net) before a crisis hits.
How we built it
We built the frontend using React and Tailwind CSS, creating a clean, accessible desktop UI that mimics the existing NatWest portal. The core intelligence is driven by Claude, which processes the speech-to-text input to map the user onto our hidden 8x8 psychological "chessboard" matrix (conceptually functioning as a mapping function $f(conversational_traits) \rightarrow (risk_tolerance, impulse_control)$). We integrated ElevenLabs to provide NatBot with a natural, empathetic voice, dynamically rendering personalized financial nudges and historical insights on the dashboard based on the user's calculated coordinate.
Challenges we ran into
One of our biggest hurdles was extracting qualitative psychological markers from a casual conversation and translating them into a quantitative financial matrix without making the user feel interrogated or categorized. Additionally, ensuring a seamless live-voice interaction meant battling API latency between the transcription, the LLM processing, and the ElevenLabs audio generation, which required us to heavily optimize our system prompts and design clever UI state transitions (like pulsing "thinking" orbs) to mask the processing time during the demo.
Accomplishments that we're proud of
We are incredibly proud of designing a fintech tool that genuinely prioritizes empathy over cold metrics. By successfully separating the complex categorization algorithm (the back-stage matrix) from the user interface (the front-stage chat), we maintained a warm, non-judgmental environment. We successfully built a working prototype that breaks down the intimidation factor of traditional banking, proving that AI can be used to foster active financial awareness and social mobility.
What we learned
We learned that impactful financial technology requires more than just accurate math; it requires behavioral psychology. By fine-tuning the LLM's persona and experimenting with voice models, we discovered how crucial tone and proactive, bite-sized nudges are when designing for users experiencing financial anxiety. We also learned how to effectively translate abstract, real-world risks (like an upcoming cold snap) into concrete, manageable financial actions.
What's next for NatAware
Moving forward, we aim to integrate real-time Open Banking APIs to process live transaction data rather than simulated histories. We also plan to expand the predictive threat engine by hooking into live localized data streams for weather, insurance risks, and public health alerts. Ultimately, we want to refine the psychological matrix to offer long-term, evolving financial coaching that seamlessly adapts as the user's social mobility and financial confidence grow.
