SmartHeritance is a Personal Legacy Manager that helps users securely organize financial accounts, digital assets, property, documents, and personal wishes in one place. SmartSync scans connected email accounts to identify assets, expenses, liabilities, and digital services, making legacy planning faster and more comprehensive across Gmail, Outlook, and Yahoo Mail.
Email-based asset discovery depends on finding a small number of meaningful messages inside large, noisy inboxes. Most personal email is marketing, alerts, or repeat notifications, while the messages that actually prove an account exists are worded inconsistently across hundreds of institutions. Turning that into a user-facing feature meant solving for speed, provider coverage, accuracy, and processing cost together.
Gmail and Outlook scans could take approximately 10 to 15 minutes before the user saw any final result. Throughout that period the interface displayed a loader with no useful progress indication, so the workflow felt stalled. The delay reduced confidence in the feature and made a core discovery step feel unreliable.
Gmail and Outlook were supported, but Yahoo Mail was not. For a legacy-planning product this was a meaningful gap, because many older users still rely on long-standing Yahoo accounts that hold years of financial, insurance, and subscription history. Those inboxes often carry the strongest evidence of forgotten accounts.
The work was approached as an email intelligence pipeline rather than a single AI prompt. Tasks that deterministic rules could handle reliably were separated from tasks that genuinely required language understanding, so each message was resolved by the cheapest dependable signal available. Speed, accuracy, and processing cost were treated as design constraints from the start.
SmartSync brought Gmail, Outlook, and Yahoo Mail into a single scan workflow. Yahoo support was added through IMAP, while provider-specific retrieval logic normalized sender, subject, message metadata, and scan state before classification. Users could discover assets from long email histories without changing providers.
The pipeline checked sender domains and duplicate signals before calling the model. A maintained domain-to-category registry recognized senders that reliably indicate an account, creating high-confidence candidates without a model call, while irrelevant and repeated messages left the expensive classification path early.
Users no longer wait 10 to 15 minutes for a single final scan response. Initial classified results now appear in roughly 10 to 12 seconds, keeping the discovery workflow interactive.
Yahoo IMAP support extended SmartSync beyond Gmail and Outlook, giving long-time Yahoo users the same automated discovery workflow without moving to a different email provider.
Domain filters, relevance checks, and duplicate suppression cut the number of emails reaching the AI layer, while deterministic matches avoided token spend on senders that already proved an account.
Eternalight built SmartSync as a staged email intelligence pipeline rather than a single AI prompt. Gmail, Microsoft Graph, and IMAP handle retrieval, rule-based domain filtering and deduplication resolve known senders, and a large language model classifies only the ambiguous financial mail. Delivery followed a structured process from brainstorming and wireframes through development, integration, launch, and user feedback.
Brainstorming
Wireframe
Development
Send us a message, and we’ll promptly discuss your project with you.
Known subscription, shopping, streaming, and cloud-service domains were passing through the full AI pipeline even when the sender domain alone gave enough evidence to identify the account. Duplicate messages triggered further model calls. The result was avoidable token usage and latency on emails that rules could resolve.
A single bank domain can send messages about credit cards, mortgages, loans, insurance, retirement products, and statements. Running every message through one broad instruction caused category confusion and raised the risk of hallucination, while expense emails and several asset categories were detected inconsistently.
Financial messages first passed through a routing layer that identified the most likely category, then were evaluated with a focused instruction for credit cards, mortgages, loans, insurance, retirement, or expenses. If the evidence did not match, the message was flagged and held for the next applicable category pass.
Mailbox retrieval was parallelized across Gmail, Outlook, and Yahoo IMAP, and independent classification jobs ran concurrently so one slow provider could not hold back unrelated results. The application polled for results and displayed classified accounts as they arrived, each carrying its category and processing state.
User Feedback
Launch
Integration
Understands and processes complex customer conversations.
Extracts insights from unstructured customer data.
Optimizes AI prompts for accurate and reliable results.
Enables secure access to Gmail emails and account data.
Supports email retrieval and asset discovery from Yahoo inboxes.
Structures and exchanges data between application components.
