Block Library (Data, Validation, NLP)
Pick the right building blocks, snap them together, and let TaxLayer run your process — your way.
Why blocks?
Each block does one job really well (e.g., “read PDFs”, “check totals”, “test in AADE sandbox”). You chain them to match your firm’s workflow. If things change (new vendor, new rule), swap a block — no rebuilds, no chaos.
You’ll get: fewer clicks, fewer errors, repeatable results, and a clean audit trail.
How the blocks fit together (at a glance)
Upload → 🧱 Data blocks → ✅ Validation blocks → 🤖 AI/Rules → 🧪 Sandbox → 🚀 Submit
Data = read & organise your documents
Validation = catch mistakes (myDATA, math, duplicates)
AI/Rules = fix safely, apply house rules, route exceptions
Sandbox = dry-run with AADE before production
📦 Data Blocks (the foundation)
Ingest & Extract
What it does: Reads PDFs/images/Excel, pulls out headers, line items, VAT, totals, dates — in Greek 🇬🇷 and English 🇬🇧. Use when: You upload a batch and need clean, structured data fast. You get: Accurate fields, confidence flags (“please review”), less manual typing.
Quick setup:
Turn Line items ON if you submit to myDATA (better VAT checks).
If scans are fuzzy, choose High OCR.
Keep files under 10MB for speed.
Pro tip: Batch similar documents together (same vendor/layout) for best accuracy.
👥 Data Management
Clients
What it does: Matches documents to the right client/vendor, standardises names, checks ΑΦΜ/VAT, avoids duplicates. Use when: Vendor names vary (“ABC LTD”, “A.B.C. Ltd”), or ΑΦΜ is missing/dirty. You get: Fewer “unknown issuer/recipient” errors and smoother submissions.
Choose which parties you wish to create your TaxLayer clients from, the issuer or the counterparty in the documents you are uploading.
Pro tip: Review your master list monthly. Clean masters = fewer downstream fixes.
✅ Validation Blocks (quality & compliance)
myDATA Rules Check
What it does: Checks document type, VAT categories, exemption codes, party details against myDATA rules. Use when: Always — it mirrors what AADE will check. You get: Early warnings on mis-classification and missing fields.
Totals & VAT Check
What it does: Confirms line totals, VAT per rate, and grand totals add up (with correct rounding). Use when: Any time you extract line items or submit to myDATA. You get: “Totals don’t add up” errors caught early — before AADE rejects.
Duplicate Detector
What it does: Spots likely duplicates by vendor + invoice number + date + amount (even with small variations). Use when: Monthly closes and bulk uploads. You get: Fewer rejections for “already submitted”.
Suggest Bulk Fixes
What it does: Groups recurring issues (e.g., missing VAT category on 18 invoices) and proposes one-click bulk fixes. Use when: You see the same error again and again. You get: Minutes instead of hours resolving clusters of issues.
myDATA Sandbox Submit
What it does: Dry-run with AADE (MARKs, error codes, the works) without touching production. Use when: New vendors, new clients, or big month-end batches. You get: Real errors translated into plain Greek/English, with fix guidance.
🧠 AI / NLP Blocks (your house rules & smart help)
Custom Review Rule
What it does: Applies your business rules in plain language. Example:
“Flag if total > €10,000 → manager approval.”
“Restaurant receipts → use expense category X.”
“If vendor on ‘new’ list → always sandbox first.”
Use when: You have firm-specific policies not covered by myDATA. You get: Consistent, documented decisions — exactly how your practice works.
Pro tip: Keep rules short and clear (“If X, then Y”). Add examples inside the rule for even better accuracy.
Friendly reminders
Start simple. A straight line beats an over-engineered maze.
Fast lane + review lane. Let clean, low-risk documents fly; slow down the rest.
Batch wisely. One client + one period = easy audit trail.
Log everything. TaxLayer does — you can export any time.
Last updated