macOS/desktop
Ai avatar app development
Create and animate 3D avatars with AI support and hardware-free motion capture feature.
AI Transaction Intelligence Platform at IPH Technologies
IPH Technologies served as the end-to-end AI and engineering partner for mEinstein, building mE Transaction — a privacy-first platform that turns scattered transactional emails into on-device financial intelligence, with an optional user-owned data marketplace layer.
Client
mEinstein (mE Transaction)
Industry
Fintech · AI · Privacy Technology
Website
meinstein.ai/finance.html
Platform
Cross-Platform Mobile App (iOS & Android) · Web Dashboard
Services Delivered
AI Architecture, On-Device ML, Backend, Mobile Engineering, Privacy Framework, Marketplace
Tech Stack
On-Device ML/NLP, OAuth 2.0, Gmail API, Microsoft Graph, Yahoo Mail & iCloud APIs, Cross-Platform Mobile, End-to-End Encryption
Turning a Buried Inbox Into Financial Clarity
Every online purchase, subscription renewal, bill payment, and banking alert adds to a person's digital financial footprint — yet most people have almost no visibility into the data sitting in their inbox.
Users keep paying for subscriptions they no longer use, miss hidden charges and refunds, and have little real control over their own financial data.
mEinstein was founded to change that by giving people ownership of their financial data through intelligence, privacy, and meaningful insight — not another app that quietly monetises what it collects.
To turn that vision into a real product, mEinstein partnered with IPH Technologies to build a platform at the intersection of AI, privacy engineering, and fintech.
Building Privacy-First Fintech AI Without Compromise
Modern consumers generate enormous financial data spread across multiple inboxes, while most solutions either ignore it or exploit it. For this project, we focused on several key engineering challenges:
01
Fragmented Financial Data — Receipts, subscriptions, invoices, and banking alerts were scattered across Gmail, Outlook, Yahoo, and iCloud, with no single source of truth or unified intelligence.
02
Zero Visibility Into Recurring Spend — Users had no reliable way of knowing which subscriptions were still active, duplicated, or had quietly increased in price.
03
Privacy as a Hard Constraint — Traditional fintech platforms process financial data in the cloud, creating centralised risk. This platform had to run AI on the device instead of in a data centre.
04
Building Trust in a Low-Trust Category — Consumer trust in fintech is low, so any architecture requiring raw data uploads was a non-starter; the engineering had to be verifiably private.
05
Monetisation Without Exploitation — Users needed to benefit financially from their own data through a mechanism that was strictly opt-in, anonymised, and fully user-controlled.
06
Scale Without Central Bottlenecks — As the user base grows, processing has to scale without driving up cloud infrastructure costs, which meant designing for edge compute from day one.
App Screens
Explore the mE Transaction App
Take a closer look at the app experience, from the unified spending dashboard to subscription tracking and the encrypted personal vault.
Features Designed for Private, Actionable Financial Intelligence
Every feature was designed around one idea: give users real financial clarity without ever asking them to hand over their raw data.
Secure Multi-Provider Email Connectivity
OAuth-based integrations connect Gmail, Outlook, Yahoo Mail, and iCloud without permanently storing credentials, with instant access revocation available at any time.
Gmail
Outlook
Yahoo & iCloud
On-Device AI Transaction Detection
An AI extraction engine identifies and classifies purchase receipts, subscriptions, utility bills, refunds, banking alerts, and more — entirely on the device, with zero raw email uploaded.
On-Device ML
15+ Categories
NLP Extraction
Subscription Tracker
Recurring charges are detected and surfaced before renewal, catching forgotten subscriptions before they silently drain an account again.
Recurring Charges
Renewal Alerts
Refund & Duplicate Payment Detection
Missed refunds and reversals are surfaced automatically, while overlapping subscriptions and double-charges are flagged before they compound.
Refund Detector
Duplicate Detection
Spending Pattern Analyser & Merchant Scorecards
Users get a clear view of merchant and category trends across every inbox, along with reliability and pricing insights on which vendors overcharge or refund quickly.
Spending Trends
Merchant Scorecards
Savings Opportunity Engine & Impulse Alerts
The platform automatically surfaces cheaper alternatives and missed discounts, while behavioural anomaly detection flags unusual spending patterns as they happen.
Savings Engine
Impulse Alerts
End-to-End Encrypted Personal Vault
All transaction intelligence is stored in a vault-based, end-to-end encrypted architecture that the user controls at every level, with nothing processed centrally.
E2E Encryption
User-Controlled
User-Owned Data Marketplace
Users can optionally contribute anonymised, tokenised transaction insights to the mEinstein Marketplace — never raw emails, never personal identities — and benefit financially from data they generate.
Opt-In Only
Anonymised
Technology Behind the Platform
The application uses an edge-native technology stack designed to support on-device AI inference, secure multi-provider email connectivity, and scalable, privacy-preserving backend services.
📱
Mobile
Cross-Platform Engineering — iOS & Android
🧠
AI / ML
On-Device ML Inference · NLP Merchant Extraction · AI Transaction Parsing
📧
Email Integration
OAuth 2.0 · Gmail API · Microsoft Graph · Yahoo Mail · iCloud APIs
🛡️
Security Architecture
End-to-End Encryption · Vault-Based Local Storage · Zero Raw-Email Monetisation
🟢
Backend Infrastructure
Scalable Backend Services · Privacy-Preserving AI Pipelines · Real-Time Analytics
💱
Marketplace Layer
Tokenised Anonymisation Framework · Decentralised Data Economy
📊
Analytics Dashboard
Spending Trends, Subscriptions & Merchant Intelligence
🔐
Authentication
OAuth-Based, Session-Scoped, No Stored Passwords
From Architecture Design to Production Deployment
We followed a structured development process, with additional focus on on-device AI engineering, secure email connectivity, and the user-owned data marketplace layer.
01
AI Architecture Design
We defined the core architectural principle — sensitive data never leaves the user's environment without explicit consent — and designed the on-device inference workflow, email connection layer, and marketplace architecture around it.
02
Secure Email Connectivity
We built OAuth-based integrations for Gmail, Outlook, Yahoo Mail, and iCloud, with token-based, session-scoped authentication and no permanently stored credentials.
03
AI Extraction Engine
We engineered the extraction engine that detects and parses receipts, subscriptions, invoices, utility bills, banking alerts, and refunds from transactional emails.
04
On-Device Intelligence
We deployed local categorisation, NLP-based merchant extraction, and trend analysis models that run entirely on-device, delivering insight generation with zero data upload.
05
Secure Personal Vault
We built the end-to-end encrypted transaction storage architecture, giving users a vault-based, fully user-controlled home for their financial intelligence.
06
Analytics Dashboard
We developed the frontend dashboard surfacing spending trends, recurring charges, the subscription tracker, and merchant intelligence in one unified view.
07
Cross-Platform Mobile Engineering
We built the cross-platform mobile application for iOS and Android, along with the backend infrastructure and scalable service layer supporting it.
08
Marketplace & Monetisation Layer
We architected the tokenised, anonymous insight-sharing marketplace layer alongside the core product, with opt-in consent and zero identity exposure built in from the start.
09
Testing & Security Validation
We tested the platform across devices and email providers, validating encryption, OAuth flows, on-device classification accuracy, and marketplace anonymisation.
10
Deployment
The iOS and Android applications and web dashboard were deployed to production, with monitoring in place to support scale as the user base grows.
What We Delivered
mE Transaction was delivered as a production-ready, privacy-first fintech AI platform with native mobile apps, a web dashboard, on-device AI, and a fully opt-in data marketplace.
4
Email providers securely integrated — Gmail, Outlook, Yahoo Mail, and iCloud
15+
Transaction categories detected and classified entirely on-device
0
Raw emails ever uploaded to a central server for processing
100%
Opt-in, anonymised participation required for the data marketplace
mE Transaction — Track It. Understand It. Monetize It.
mE Transaction is a privacy-first AI platform that turns transactional emails into meaningful financial insight — without ever uploading a single one. IPH Technologies handled the project end-to-end, from architecture and privacy engineering through mobile, backend, and marketplace development.
Fintech AI // Transaction Intelligence Platform
mE Transaction.
Your Data.
Your Rules.
Built for mEinstein, the platform connects securely to Gmail, Outlook, Yahoo, and iCloud, classifies transactions on-device across 15+ categories, and gives users the option to monetise anonymised insights through a fully consent-based marketplace.
On-Device ML
OAuth 2.0
Gmail API
Microsoft Graph
iOS
Android
End-to-End Encryption
Frequently Asked Questions
IPH Technologies designed and built mE Transaction end-to-end — covering AI architecture, on-device ML, secure email integrations, iOS and Android apps, backend infrastructure, analytics dashboard, privacy framework, and the marketplace layer, as one partner delivering the full stack.
Privacy was treated as the first design decision, not a final checkbox. Every technical choice — encryption, OAuth token scoping, local inference, data minimisation — was evaluated against a single principle: sensitive financial data should never leave the user's device without explicit consent.
On-device AI platform development runs inference directly on the user's phone instead of a central server. For mE Transaction, NLP-based merchant extraction, transaction classification, and insight generation all happen locally, eliminating centralised data risk and reducing cloud infrastructure costs.
The platform integrates with Gmail, Microsoft Outlook, Yahoo Mail, and Apple iCloud through OAuth 2.0. No credentials are permanently stored, only transactional emails are analysed, and users can revoke access instantly through in-app controls.
The platform relies on end-to-end encryption, OAuth-based authentication with no password storage, vault-based local storage, and strict data minimisation at the extraction layer — an architecture designed to be auditable by enterprise compliance teams and transparent to end users.
The AI extraction engine automatically identifies subscription renewals, utility bills, and other recurring transactions, surfacing them through a subscription tracker before renewal so users can catch forgotten or duplicated charges before they compound.
Yes. Users can optionally contribute anonymised, tokenised transaction insights to the mEinstein Marketplace — never raw emails or personal identities — with full opt-in consent required at every step.
Yes. The on-device AI architecture, secure data handling framework, and user-owned data economy layer built for mE Transaction apply across healthcare, legal, insurance, and any domain where sensitive data processing and user trust are non-negotiable.
.png)
