Data Readiness & Privacy

Garbage in, garbage out

AI can't fix messy notebooks, conflicting prices in different chats, or customer lists with no phone numbers. Clean, consistent data makes every AI use case better — from chat replies to sales forecasts.

Spend a week organizing before you spend money on fancy tools.

Minimum data hygiene checklist

• check_circleOne source of truth for prices, hours, and menu/product list
• check_circleCustomer contacts in one place (CRM, Sheet, or WhatsApp Business labels)
• check_circleNamed files — "Menu_2026.pdf" not "final_FINAL2.docx"
• check_circleRemove duplicate rows and fix obvious errors in spreadsheets
• check_circleDocument who owns updating each dataset

What never goes into consumer AI tools

Use business/team plans with data protection terms when customer information is involved. When in doubt, anonymize: "Customer A ordered X" instead of full names and IDs.

• check_circleNational ID numbers, passport scans, full bank or MoMo PINs
• check_circleUnredacted medical or salary records
• check_circleClient contracts marked confidential without business-tier privacy terms
• check_circlePasswords, API keys, or admin credentials

Customer trust in Rwanda

Rwandan customers trust people, not bots hiding them. Be transparent if AI helps draft messages. Never let AI initiate payments or promise refunds without human approval.

If you store customer data in apps or databases, know where servers are hosted and who can access exports.

Prepare data for RAG and custom bots

Retrieval-augmented generation (RAG) means AI answers from your documents — FAQs, policies, product catalogs. Those documents must be accurate and updated when prices or hours change.

A public FAQ page on your website is often the best first "knowledge base" — free, customer-facing, and easy for AI tools to reference.