SANDI — AI Card Savings Concierge

What the company appears to be: SANDI is a mobile-first, location-aware AI concierge that helps users determine which credit card to use for a given purchase by analyzing existing cards in their wallet and suggesting savings opportunities. It operates as a prototype built during OpenAI Build Week, using a hybrid AI-and-rules system. The product does not connect to bank accounts or submit card applications.

What changed: The project began with the frustration of complex credit card reward optimization and evolved into a working prototype that uses natural language input, location awareness, and structured financial logic to recommend cards based on user spending patterns and existing card benefits.

Single most important open question: Does SANDI have any evidence of traction, revenue, or customer adoption beyond its prototype status?

The description states that SANDI is a mobile-first, location-aware AI card-savings concierge. It allows users to describe purchases in everyday language and then:

• Extracts purchase details (category, merchant, location, amount, timing)
• Searches for nearby places and eligible offers
• Compares existing cards in the user's wallet
• Explains the best option, estimated savings, thresholds, and alternatives
• Switches between text explanation and map with reward-aware merchant markers
• Separates "best card I already own" from "possible new-card opportunity"
• Helps plan larger future purchases when timing may matter
• Tracks benefits, reminders, and upcoming spending plans

The public build is described as a working prototype using sample data. It does not connect to bank accounts, submit card applications, determine credit eligibility, or complete reward redemption.

Evidence: The author's own write-up describes these features in detail.

Inference: This appears to be an early-stage product concept focused on improving user experience around credit card usage through AI and location intelligence.

The description states that SANDI began with a frustration: "earning the best credit-card reward should not require five bank pages, a spreadsheet, and mental math at checkout."

It positions itself as reversing the logic of most comparison tools by helping people get more value from cards they already have before suggesting new ones.

Evidence: The author's own write-up describes this evolution from frustration to solution.

Inference: This suggests a positioning shift from generic credit card tools toward a more personalized, user-centric approach that emphasizes existing asset optimization over acquisition.

The description does not provide explicit information about target customers or ideal customer profiles (ICP). It only mentions that users describe purchases in everyday language and that the tool is designed for people who want to optimize their card rewards.

Evidence: Not evidenced.

Inference: Based on the product's focus, it likely targets individuals with multiple credit cards who are interested in maximizing rewards but find current tools too complex or time-consuming.

There is no evidence provided about business model or pricing. The description states that the public build is a prototype and does not connect to bank accounts, submit card applications, determine credit eligibility, or complete reward redemption.

Evidence: Not evidenced.

Inference: If this were to become a commercial product, it would likely involve partnerships with banks or card issuers, possibly through a freemium model or subscription-based access. However, no such details are stated.

The system is described as a hybrid AI-and-rules system:

• Mobile interface uses React 18, JavaScript, Babel, plain CSS, Leaflet, and MapLibre
• Backend uses Python and FastAPI with SQLite and RTree spatial indexing
• Uses GPT-5.6 via OpenAI Responses API for structured intent extraction
• Falls back to local rule-based parser when AI is unavailable
• Data layer indexes over 300,000 urban points of interest
• Supports wallet management, text and browser-based input, recommendation results, nearby exploration, future-purchase planning, and a text-to-map answer mode

Evidence: The author's own write-up describes the technical architecture.

Inference: This indicates a strong engineering foundation with both AI integration and deterministic logic for financial calculations. The use of Codex suggests rapid iteration capabilities.

The description states that this is a "working prototype" built during OpenAI Build Week. It does not mention any revenue, customers, or adoption metrics beyond its prototype status.

Evidence: Not evidenced.

Inference: As a hackathon submission, it has no traction data. The project is at an early stage of development and lacks real-world usage or market validation.

The description does not provide information about competitors or competitive landscape.

Evidence: Not evidenced.

Inference: Given its focus on optimizing existing card rewards rather than promoting new cards, it may compete with general credit card comparison tools or apps that help users track rewards. However, no specific competitor names or market positioning are mentioned.

Key risks and red flags include:

1. Prototype-only status: No evidence of real-world usage or customer adoption.
2. No financial data integration: Cannot connect to actual bank accounts or submit applications.
3. Limited scope: Currently focused on a small set of categories (e.g., dining) in one city (New York).
4. AI dependency: Relies heavily on GPT-5.6 for intent parsing, which could be unreliable or unavailable.
5. Compliance and trust issues: The prototype does not include secure authentication, consent controls, or encrypted storage—key requirements for financial products.
6. Lack of monetization strategy: No indication of how the product will generate revenue.

Evidence: These are inferred from the lack of traction data, technical limitations, and absence of business model details.

1. What is the current stage of development beyond the prototype?
2. Have you identified potential partners or issuers willing to collaborate?
3. How do you plan to address compliance requirements for handling financial data?
4. What are your plans for scaling beyond New York City and specific spending categories?
5. Is there any interest from banks or credit card companies in piloting this product?
6. How will you measure success once launched—user engagement, savings achieved, or other KPIs?