Inspiration
Subscriptions are convenient to start but difficult to manage. Free trials quietly become paid plans, prices increase without being noticed, and renewal reminders are buried across hundreds of emails.
For individuals and families using streaming services, software tools, memberships, cloud storage, and AI products, forgotten subscriptions can add up to hundreds or even thousands of dollars each year.
We were also concerned about phishing emails disguised as invoices, renewal notices, or cancellation messages. This inspired us to create SubGuard: AI Subscription Guardian—a tool that helps users understand recurring expenses while identifying suspicious subscription-related emails.
What it does
SubGuard connects with Apple Mail on macOS and analyzes subscription-related messages from the user’s existing mailbox.
It can identify:
- Merchant or service name
- Subscription price
- Monthly or annual billing cycle
- Free-trial expiration date
- Upcoming renewal date
- Price increases
- Duplicate or overlapping subscriptions
- Estimated monthly and annual spending
- Potential savings
- Suspicious senders, domains, and cancellation links
The extracted information is organized into a clear dashboard. Each subscription receives an actionable recommendation such as Keep, Review, Cancel, or Security Warning.
Users can analyze their existing Apple Mail messages instead of manually copying every email into the application.
How we built it
We built SubGuard as a local-first macOS subscription analysis application.
The project integrates with Apple Mail, allowing the system to access and analyze relevant messages from the mailbox already configured on the user’s Mac. This approach lets users work with Gmail, Outlook, iCloud Mail, and other accounts that are already connected through Apple Mail, without requiring us to build a separate integration for every email provider.
The application uses AI-based extraction to convert unstructured email content into structured subscription records. Rule-based validation is then used to check dates, prices, sender information, domains, suspicious wording, and potentially unsafe links.
The extracted records are used to calculate estimated monthly costs, annual costs, upcoming renewals, and potential savings.
Codex helped us plan the workflow, implement the Apple Mail integration, develop the analysis pipeline, debug the application, and test the system.
Challenges we ran into
One major challenge was that subscription emails do not follow a standard format. Important information may appear in the subject line, body text, receipt tables, disclaimers, or attached documents.
Another challenge was working with Apple Mail data reliably. Messages may contain HTML, plain text, quoted replies, embedded images, unusual character encodings, and large attachments. We needed to extract the useful content while ignoring signatures, tracking information, and irrelevant formatting.
It was also difficult to distinguish between:
- A real subscription notice
- A promotional email
- A payment receipt
- A free-trial reminder
- A phishing message pretending to be a billing notice
Pricing information created additional ambiguity because one email may contain discounts, taxes, previous prices, future prices, and one-time charges.
We also had to design the Apple Mail connection carefully so the application could provide useful analysis while keeping email access transparent and under the user’s control.
Accomplishments that we're proud of
We successfully connected SubGuard to the native Apple Mail application on macOS and used real mailbox data as part of the subscription analysis workflow.
This means users do not need to reconnect every Gmail, Outlook, or iCloud account separately. If an account already works inside Apple Mail, SubGuard can use Apple Mail as a unified local email interface.
We are also proud that SubGuard combines two normally separate problems—subscription management and email security—into one understandable experience.
The project does more than list recurring expenses. It explains why an email matters, what financial or security risk it may represent, and what action the user should consider next.
Our core workflow is:
- Connect to Apple Mail
- Locate relevant subscription and billing messages
- Extract prices, renewal dates, and service information
- Evaluate financial and security risks
- Present clear recommendations and potential savings
What we learned
We learned that Apple Mail can act as a useful unified access layer for multiple email providers. Rather than creating separate integrations for Gmail, Outlook, and iCloud, we can work with accounts the user has already configured on their Mac.
We also learned that email contains more financial context than a standard bank transaction.
A bank statement may show that a company charged $19.99, but the related email may explain what the service includes, whether it began as a free trial, when the next renewal will happen, whether the price is changing, and how the account can be managed.
AI extraction works best when combined with explicit validation rules. AI is effective at understanding language and context, while deterministic checks provide more consistent results for domains, URLs, dates, prices, and security indicators.
Most importantly, users do not only need more financial data. They need concise explanations and clear actions.
What's next for SubGuard: AI Subscription Guardian
Our next step is to improve the Apple Mail analysis workflow so SubGuard can process larger mailboxes faster and identify subscription messages with greater accuracy.
Future development may include:
- Automatic subscription discovery across Apple Mail folders
- Receipt and PDF attachment analysis
- Renewal and trial-expiration reminders
- Family and team subscription dashboards
- Bank transaction matching
- Usage-based subscription recommendations
- Multilingual email analysis
- Local AI model support
- Guided cancellation workflows
- Subscription price-change history
- Duplicate and shared service detection
- Windows and web-based email integrations
Our long-term goal is to make SubGuard an AI-powered financial safety layer that helps users understand recurring expenses, avoid unnecessary renewals, and identify suspicious billing messages before they lose money.
Built With
- apple-mail
- applescript
- macos
- openai-codex
- python
Log in or sign up for Devpost to join the conversation.