CodeIQ FAQ

CodeIQ is an intelligent bookkeeping assistant that automates the coding of bank transactions. It suggests the correct nominal codes, VAT rates, and descriptions based on advanced pattern learning and semantic analysis.

The platform integrates directly with your accounting software (Pandle, QuickBooks, Xero, Sage) and processes bank statements through an intelligent processing pipeline that includes:

• Transfer detection to prevent double-entry
• Invoice matching for automatic payment posting
• Historical pattern recognition from your own coding history
• Universal pattern database with crowd-sourced intelligence
• Semantic intelligent coding using fine-tuned embeddings
• VAT classification and compliance

You simply review the suggestions and approve them - CodeIQ handles the posting to your accounting platform automatically.

CodeIQ is designed for accounting practices managing multiple clients. You can open up to 10 concurrent client sessions in a single browser window, each with its own tab.

The multi-session architecture allows you to:

• Process different clients in parallel without waiting
• Switch between client tabs instantly with no reloading
• Run background processing for multiple clients simultaneously
• Maintain separate session states, chart of accounts, and VAT codes per client

The interface uses pagination to display 8 visual tabs at a time, making it easy to manage even 10+ active sessions.

CodeIQ currently integrates with four major UK accounting platforms:

• Pandle - Primary integration with deepest feature support
• QuickBooks Online - Full GL history and pattern contribution
• Xero - Bank transaction learning and pattern contribution
• Sage - CSV-based processing with pattern contribution

Each platform integration uses OAuth 2.0 for secure authentication. CodeIQ automatically fetches your chart of accounts, VAT codes, and outstanding invoices to provide intelligent coding suggestions.

CodeIQ uses a sophisticated multi-tier learning system:

1. User Pattern Learning: Every time you approve or correct a transaction, CodeIQ stores that pattern in your private user pattern database. Next time it encounters a similar merchant or transaction description, it will recall your preference and apply it automatically.

2. GL History Analysis: When you first connect your accounting platform, CodeIQ reads your general ledger transaction history to learn how you've historically coded transactions. This builds your baseline pattern cache.

3. User Pattern Overrides: Your personal coding rules always take priority. If you've explicitly coded a merchant a certain way, that overrides even universal patterns.

The system continuously learns and improves as you use it, becoming more accurate over time.

The Universal Pattern Database is CodeIQ's collective intelligence system - a crowd-sourced database of thousands of transaction patterns contributed by users across the platform.

How it works:

• CodeIQ analyzes your GL history for high-quality patterns (merchants you've coded consistently)
• Patterns are anonymized and PII is removed
• Submitted patterns go to a staging area for admin approval
• Approved patterns are cleaned by LLM for consistency
• Cleaned patterns are added to the universal database available to all users

The Network Effect: As more users contribute, the database becomes smarter. When you encounter a merchant for the first time, there's a high chance CodeIQ already knows the correct coding from other users.

This is the most heavily used matching method in CodeIQ's processing pipeline, providing accurate suggestions even for transactions you've never seen before.

Semantic Analysis uses CodeIQ's advanced semantic model to understand the meaning and context of transactions, not just pattern matching.

Key advantages:

• Cost-effective: Runs on our infrastructure with no per-transaction API costs
• Context-aware: Understands semantic similarity between descriptions (e.g., "office supplies" and "stationery")
• Handles variations: Works even when merchants use different naming conventions
• Industry-specific: Fine-tuned on accounting terminology and UK business transactions

This acts as a fallback when pattern matching doesn't find an exact match, providing intelligent suggestions based on semantic understanding.

Step 1: Connect & Initialize

• Authenticate with your accounting platform via OAuth
• Select target company/client
• CodeIQ fetches chart of accounts, VAT codes, and bank accounts

Step 2: Upload Bank Statement

• Upload CSV, PDF, Excel, OFX, or QBO file
• CodeIQ auto-detects columns (date, description, amounts)
• Confirm or adjust column mappings

Step 3: Processing (Background)

• Transfer detection identifies inter-account transfers
• Invoice matching links payments to outstanding invoices
• Pattern learning applies historical and universal patterns
• Semantic analysis provides intelligent coding suggestions
• VAT classification determines correct tax rates

Step 4: Review & Approve

• Review suggestions with confidence scores
• Make corrections where needed (CodeIQ learns from these)
• Approve batches of transactions

Step 5: Automated Posting

• CodeIQ posts approved transactions to your accounting platform
• Bank accounts are automatically reconciled
• Real-time progress tracking via posting center

CodeIQ accepts a wide range of bank statement formats:

• CSV - Comma-separated values (most common)
• Excel - .xlsx and .xls formats
• PDF - Parsed using custom bank parsers
• OFX - Open Financial Exchange format
• QBO - QuickBooks export format

The CSV upload includes intelligent column detection that automatically identifies date, description, debit/credit, and balance columns. You can confirm or adjust the mappings before processing.

Transfer detection prevents double-entry errors by identifying transactions that move money between your own bank accounts.

CodeIQ analyzes transactions for:

• Equal and opposite amounts within a short time window
• Matching descriptions indicating transfers
• Same-day or next-day clearing patterns

When a transfer is detected, CodeIQ pairs the transactions and prevents them from being double-coded as income/expense. Instead, they're properly posted as inter-account transfers.

Invoice matching automatically links bank transactions to outstanding invoices in your accounting system, enabling one-click payment posting.

CodeIQ fetches your platform's outstanding invoices and matches them to bank transactions based on:

• Amount matching (exact or within tolerance)
• Customer/supplier name matching
• Invoice reference number in transaction description
• Date proximity to invoice due date

When a match is found, you can approve it and CodeIQ will:

• Post the payment to the accounting platform
• Mark the invoice as paid
• Reconcile the bank account

This is especially powerful for businesses with high invoice volumes.

Invoice OCR Recognition lets you drag and drop invoices directly into CodeIQ, automatically extracting key data without manual entry. It works like Dext or AutoEntry, but built directly into your bookkeeping workflow.

Supported file formats:

• PDF invoices (up to 10MB)
• PNG images
• JPG/JPEG images

Extracted fields:

• Invoice number and date
• Contact/supplier name
• Gross amount, net amount, and VAT
• Currency (GBP, USD, EUR, and more)
• Line item descriptions
• Suggested expense/income category

How it works:

1. Drag and drop one or multiple invoice files into the upload area
2. Gemini Vision processes each document and extracts data
3. Review extracted data with confidence scores for each field
4. Invoices are cached and can be matched to bank transactions
5. When a matching payment appears, post both in one click

All uploaded files are encrypted with AES-256-GCM during transit and processing.

Yes. CodeIQ implements enterprise-grade security measures:

• OAuth 2.0 authentication: We never store your accounting platform credentials
• Encrypted connections: All data transfer uses HTTPS/TLS
• Session-based processing: Transaction data is session-scoped and expires after 24 hours
• Local embeddings: Semantic model runs on our server, no third-party APIs
• PII anonymization: Pattern contributions have personal information stripped before submission
• Isolated data: Each user's pattern learning is private and never shared without permission

We do not train public models on your private client data unless you explicitly opt in to pattern contribution.

Pattern contribution is the system that allows CodeIQ to learn from your GL history and contribute anonymized patterns to the universal database, benefiting all users.

What's contributed:

• Merchant names (cleaned and normalized)
• Transaction descriptions (PII removed)
• Account codes and names
• VAT rates
• Occurrence frequency

What's NOT contributed:

• Transaction amounts
• Dates
• Client/company names
• Personal information
• Bank account details

All patterns go through admin review before being added to the universal database. You can opt out of pattern contribution in your account settings, though this means you won't benefit from the collective intelligence as much.

CodeIQ uses a credit-based pricing model where each transaction processed consumes one credit. Pricing tiers are designed for different business sizes:

• Starter: 500 credits/month - suitable for freelancers or small businesses
• Professional: 2,500 credits/month - for accountants with multiple clients
• Practice: 10,000 credits/month - for accounting practices
• Enterprise: Custom pricing for high-volume users

Credits renew monthly and unused credits don't roll over. You can also purchase one-time credit top-ups if you exceed your monthly allocation.

Yes! New users receive 500 free credits to test the platform. This is enough to process a typical month's bank statement for a small business.

We also offer a 14-day money-back guarantee on all paid plans if you're not satisfied with the results.

If you exceed your monthly credits, you can purchase one-time credit top-ups immediately to continue processing. Alternatively, upgrade to a higher tier plan for more monthly credits.

Enterprise plans offer custom credit limits for high-volume users.

CodeIQ shows confidence scores for each suggestion, allowing you to prioritize review of low-confidence items. Nothing is posted without your approval.

If you notice an error:

• Simply correct it in the review interface
• CodeIQ immediately learns from your correction
• Future similar transactions will use your corrected coding

The system becomes more accurate the more you use it, learning your specific coding preferences over time.

No. CodeIQ is a fully cloud-based platform that runs entirely in your web browser. There's no software to download, install, or maintain.

Benefits of the cloud-based approach:

• Access from any device with an internet connection
• No compatibility issues with different operating systems
• Always use the latest version with automatic updates
• Work from home, office, or client sites seamlessly

Simply log in through your web browser and start processing transactions.

CodeIQ works on all modern browsers:

• Chrome (recommended)
• Firefox
• Safari
• Microsoft Edge

We recommend using the latest version of your preferred browser for the best experience.