Sentra - Personal Finance Sonar

Inspiration

Do you know the recent policy passed on student loan forgiveness? How about the latest tax bracket changes? Policies, laws, and economic events shift every week, and most people aren’t aware of how these changes impact their personal finances. Many miss opportunities or make poor decisions- not because they don’t care, but because staying informed is overwhelming. Sentra was born to bridge that gap: to passively and proactively inform everyone-regardless of background-how the world is affecting their money.

What it does

Sentra is a smart financial companion that connects to your financial accounts (via Plaid) and continuously scans real-world events using the Perplexity Sonar model. From inflation data and interest rate hikes to tax code revisions and regulatory updates, Sonar ingests global information in real time—and Sentra personalizes it to you.

By combining financial data with cutting-edge AI reasoning, Sentra translates complex news into concise, context-aware insights tailored to your unique financial profile. The result: a constantly updating radar that lets you anticipate and respond to financial change-without needing to track headlines or understand economics.

How we built it

• User Research: I began with real conversations-friends and family shared how hard it is to stay informed and confident in financial decisions. This guided the design toward real-world utility and accessibility.

• Backend: Using Plaid, I securely aggregated users' financial data (checking, credit, investments, loans). Then came the heart of the project-integrating Perplexity Sonar. Sonar continuously interprets global economic signals, policies, and market events, and I built a system that translates Sonar’s outputs into personal impact assessments.

• Prompt Engineering & Insight Mapping: To get precise, high-impact results from Sonar, I developed a dynamic prompt pipeline. This wasn't just about summarizing news-it was about teaching the model to reason through the implications of world events for real people’s finances. I built logic to map each Sonar insight back to users' data, turning a generic headline into a personal financial signal.

• Frontend: I prototyped a clean, insight-first UI focused on financial clarity. Every design decision prioritized readability, actionability, and approachability-especially for users without financial expertise.

• Iterative Testing: I stress-tested every layer-Sonar queries, mapping logic, user flows-to ensure that the insights users see are not just accurate, but truly relevant and timely.

Challenges we ran into

• Taming the Firehose: The global information stream is massive. Filtering it into insights that are both personally relevant and trustworthy required deep integration between Sonar and the user’s financial context.

• AI Prompt Engineering: Sonar is powerful, but it requires precision. I had to master how to guide it—structuring prompts, fine-tuning context, and aligning outputs with what real users need.

• Trust & Security: With Plaid integration, data privacy was non-negotiable. I prioritized encrypted connections, secure handling, and user-centric transparency throughout.

• Designing for All: Financial literacy varies widely. The biggest design challenge was creating an interface that feels empowering-not overwhelming-regardless of the user's experience with finance.

Accomplishments that we're proud of

• AI-First Insight Engine: Created a robust pipeline that uses Perplexity Sonar as a reasoning layer, not just a summarizer-bridging the gap between raw news and personal financial advice.

• True Personalization: Successfully matched macro events to micro financial data—if inflation is hitting food costs or a new policy affects loan interest, Sentra surfaces it when it matters to you.

• Empowering Design: Built a product that simplifies complexity, turning economic chaos into financial clarity for everyday users.

What we learned

• Sonar’s Power Is in Reasoning: It’s not just about scraping headlines—it’s about making sense of them. Perplexity Sonar brought a new level of contextual reasoning that would be nearly impossible to hard-code.

• Prompting is Product Design: Designing the right AI prompts is as critical as UI design. Every Sonar interaction had to be shaped, iterated, and tested to produce the right kind of intelligence.

• The UX of Trust: Delivering insight is only useful if the user understands and trusts it. The biggest lesson was balancing intelligence with usability.

What's next for Sentra – Personal Finance Sonar

• Richer Sonar Contexts: Explore deeper economic models and integrate more diverse sources into Sonar’s inputs to expand the scope and granularity of insights.

• Proactive Impact Alerts: Build real-time alerting when Sonar detects events that are about to impact a user’s finances—before they feel it in their bank account.

• AI-Driven Financial Coaching: Combine Sonar’s insights with tailored educational modules and action steps—so users not only know what’s happening, but also what they can do.

• Community + Feedback Loop: Empower users to respond to insights, contribute context, and help the AI learn what matters most.

Built With

• javascript
• plaid
• postgresql
• python
• react
• solar-perplexity