Making business documents more useful.
MyLedgerLens helps businesses transform unstructured financial paperwork into organized, verifiable information. By combining AI-assisted extraction with structured validation and human review, our vision is to make document-heavy workflows simpler and more reliable.
THE PRACTICAL CONTEXT
Important business information should not be trapped in paperwork.
In almost every industry, essential operating details are distributed across PDFs, scanned receipts, printed delivery manifests, and complex supplier paperwork.
Before a finance team can balance a ledger, check a price variance, or confirm an inventory receipt, someone must manually read each document, find the key figures, and re-type them into software.
This manual transcription is not only tedious; it is inherently fragile. Typography shifts, columns run together, vendor formats differ from one supplier to the next, and simple typing errors introduce subtle accounting discrepancies that take days to reconcile at the end of the month.
We believe that modern artificial intelligence, applied with precision and rigorous validation, can fundamentally resolve this friction by turning visual document structures into clean, queryable data.
PRODUCT & TECHNICAL ARCHITECTURE
The technology behind MyLedgerLens
MyLedgerLens processes documents through a multi-stage architecture that combines semantic visual extraction, deterministic arithmetic verification, and transparent human oversight.
AI-powered extraction
MyLedgerLens uses Google Gemini 2.5 Flash to interpret supported invoice documents and extract structured information, including supplier details, invoice numbers, dates, line items, taxes, and totals.
Deterministic validation
The platform checks extracted data against defined rules, including arithmetic consistency and date-related issues. The system flags discrepancies for review rather than silently changing source information.
Human review
Invoices requiring attention can be reviewed, corrected, approved, or rejected through the existing review workflow, ensuring human oversight and accountability over financial records.
Structured data and exports
Approved invoice data can be organized for downstream business use, with CSV export available where supported by the existing implementation.
Secure document handling
Document storage and relational records are managed through Supabase with workspace-scoped Row-Level Security (RLS) and encrypted object storage, restricting access strictly to authenticated workspace members.
No silent modifications
Extracted values are never silently guessed or modified. Every document retains an immutable original copy, a structured JSON extraction payload, and an explicit log of all arithmetic and consistency checks.
CURRENT STATUS: WORKING MVP
From document upload to review-ready data
MyLedgerLens is at the working MVP stage, with an invoice-processing workflow designed to extract document data, identify potential inconsistencies, and support human review before approval.
Upload an invoice
Upload supported invoice documents (PDF, PNG, JPEG) directly into encrypted workspace storage.
Extract structured data with AI
Google Gemini 2.5 Flash interprets document layout and extracts supplier details, dates, line items, and totals.
Validate fields and calculations
Deterministic arithmetic checks verify subtotal, taxes, and grand totals with a strict 2-cent rounding tolerance.
Review flagged issues
Calculated discrepancies, missing mandatory fields, and historical workspace duplicates are surfaced for review.
Approve or reject the invoice
Human operators inspect side-by-side document views, correct fields if needed, and confirm approvals.
Export supported structured data
Approved records export into standardized CSV files with active protection against spreadsheet formula injection.
Implemented Scope: Live workflow covers single-invoice upload, Gemini multimodal extraction, arithmetic verification, duplicate detection, human review screens, and CSV export.
VERIFIED IMPLEMENTATIONEmpirical Benchmark & Active Pilot Traction
The team behind MyLedgerLens
Combining hands-on software engineering, multimodal AI systems, and cloud infrastructure to make document-heavy accounts payable workflows reliable, auditable, and automated.
David Osili
Founder & Lead Architect
Leads core platform engineering, deterministic validation logic, and end-to-end technical architecture across MyLedgerLens.
Tariq Adeleke
Chief Technology Officer (CTO)
Directs multimodal vision pipelines, OCR document layout parsing, and specialized pre-built Azure AI Document Intelligence model integration.
Emmanuel Nnamdi
VP of Engineering
Heads enterprise cloud reliability, multi-tenant security boundaries, high-throughput batch ingestion, and scalable Azure cloud infrastructure.
Folashade Balogun
Chief Operating Officer (COO)
Oversees design-partner pilot deployments, accounts payable workflow design, and commercial partnerships across wholesale distribution and accounting practices.
"Our mission is to solve the most error-prone bottleneck in accounting: transforming messy paperwork into clean, auditable ledger data without guessing or silent alterations. We build every feature around deterministic arithmetic checks and human operator oversight."
— David Osili, Founder & Lead Architect
Deterministic Math Rules
Strict arithmetic verification on subtotals, statutory tax lines, and grand totals with a 2-cent boundary.
Operator Accountability
Full side-by-side inspection and tamper-proof relational audit logs on all document approvals.
INFRASTRUCTURE & CLOUD ARCHITECTURE
Cloud infrastructure and enterprise scale roadmap
MyLedgerLens combines an active working MVP foundation with a structured enterprise roadmap designed to scale high-throughput extraction workloads natively on Microsoft Azure.
Multimodal AI Processing
Powered by Google Gemini 2.5 Flash for high-speed document understanding, vision processing, and structured schema extraction during initial MVP validation.
PostgreSQL & Object Storage
Built on Supabase managed infrastructure with multi-tenant Row-Level Security (RLS) policies and encrypted document bucket isolation.
Serverless Edge Deployment
Deployed via Vercel edge infrastructure with serverless execution boundaries, automated branch builds, and strict security headers.
Target Microsoft Azure Architecture & AI Growth Plan
As MyLedgerLens transitions from pilot validation to high-volume commercial accounts-payable workflows, our architecture is engineered to run compute-intensive workloads natively on Microsoft Azure's enterprise AI and cloud infrastructure.
Azure AI Document Intelligence
Targeted for enterprise OCR, layout analysis, and specialized pre-built invoice and receipt extraction models. Delivers pixel-accurate bounding box coordinates and robust tabular parsing across dense, variable-format supplier manifests.
Azure OpenAI Service
Deploying private Azure OpenAI GPT-4o models for complex line-item cross-examination, semantic discrepancy explanations, and multi-lingual bill reconciliation—backed by strict enterprise data governance and zero model training on customer payloads.
Azure Blob Storage & App Services
Dedicated document isolation with Customer-Managed Keys (CMK), immutable compliance audit logs, and containerized Azure App Services / Azure Container Apps for elastic, horizontal batch throughput with sovereign regional data residency.
Architecture Portability & Microsoft Azure Commitment
As MyLedgerLens scales toward commercial availability across wholesale distribution networks and independent bookkeeping firms, Microsoft Azure serves as our primary target enterprise cloud ecosystem. Transitioning high-throughput workloads to Azure AI Document Intelligence and Azure OpenAI enables enterprise SLA guarantees, strict regulatory data residency, and seamless accounts-payable integration for institutional clients.
EXPERIENCE THE WORKFLOW
See how MyLedgerLens works
Explore the working MVP to see how intelligent extraction, deterministic checks, and human review come together in an active document workflow.