Brain Knowledge Setup (9-step wizard)
Turn your AI from “generic” into “your business expert” in 15 minutes.
🚀 Why This Matters
Think of the Brain Knowledge Setup as teaching your AI how you work. Without it, the AI is a smart generalist. With it, the AI becomes your in-house myDATA expert — fluent in your accounting system, your vendors, your error patterns, and your compliance rules.
What changes after setup?
✅ Accuracy: 95%+ (instead of “almost right” answers)
✅ Errors: 80–90% fewer classification mistakes
✅ Manual work: Cut by 60–75%
✅ Speed: Month-end closes 3–5x faster
🔎 Pain point solved: No more “AI almost got it right, but I still had to fix everything.”
🧭 The 9 Steps (Quick Overview)
To access the Brain AI knowledge training form simply click on the fill knowledge button in the Brain AI page.
1️⃣ Accounting Software Details
Tell us which system you use (e.g., Xero, QuickBooks, SAP).
Why?
Each system formats fields differently. This ensures the AI maps invoices correctly and speaks your system’s “language.”
2️⃣ myDATA Preferences
Decide how cautious you want to be: sandbox only, manual reviews, or auto-submit.
Why?
You’re always in control — start safe, then add more automation as trust builds.
3️⃣ Document Types
Select which documents you handle most: invoices, receipts, credit notes, expenses.
Why?
The AI focuses its accuracy on the 80% you process daily.
4️⃣ Monthly Volumes
Estimate how many docs you process in a typical vs. peak month.
Why?
This tunes performance — -- TaxLayer applies different rules for high volume vs low volume transactions to optimise speed.
5️⃣ Common Error Patterns
Share your “usual suspects” (e.g., wrong VAT category, missing totals).
Why?
The AI learns to spot these before they trip you up.
6️⃣ Routine Automations
Tell us which repetitive tasks eat up most of your time:
Data entry
Invoice matching
Expense categorisation
VAT calculations
Why?
The AI learns your workflow priorities and builds automation around your biggest time sinks.
7️⃣ Automation Preferences
Choose your comfort level:
🟢 Safe fixes → auto-apply
💡 Suggestions → need approval
📦 Bulk → apply to many docs at once
Why? Balance speed with control — and adjust anytime.
8️⃣ Historical Data
Upload 10–20 “gold standard” examples (correct invoices, vendor layouts, categories).
Why?
Your AI learns your patterns instead of guessing.
9️⃣ Outcome Goals
Rank what matters most: accuracy, speed, compliance, or automation.
Why?
The AI prioritises based on your goals — not just generic defaults.
📈 What You’ll Notice After Setup
Day 1:
Extraction accuracy improves instantly
Validation checks match your business
Auto-fixes clear common errors automatically
Week 1:
Error rates drop sharply
Vendor-specific corrections kick in
Batch reviews take minutes, not hours
Month 1:
AI feels like a member of your team
Manual corrections minimal
Predictive error prevention in place
🔒 Privacy (Plain English)
Your AI “brain” is private — it’s only for your firm.
Your invoices are never used to train global models.
Everything is encrypted, logged, and fully GDPR-compliant.
💡 Pro Tips
Bring 10–20 perfect sample documents — they teach the AI better than any setting.
Start conservative → sandbox + manual review → then move toward automation.
Tell the AI about your weird cases (edge cases save hours later).
Revisit setup quarterly — your business evolves, your AI should too.
👉 In short: spend 15 minutes setting up Brain Knowledge now, and save dozens of hours every month after. It’s the foundation that makes the AI genuinely yours.
Last updated