AI Features

What are Tawridy's AI Features?

Tawridy embeds artificial intelligence across the entire procurement lifecycle to reduce manual data entry, surface actionable insights, and catch potential risks before they become problems. From parsing unstructured inquiry emails into structured line items to detecting suspiciously low prices that may indicate counterfeit goods, AI works as an intelligent assistant at every stage. Critically, all AI features are optional suggestions that augment your team's judgment. They never override human decisions or execute actions automatically without confirmation. You stay in control while AI handles the tedious work.

How It Connects

Direction	Module	Relationship
In	RFQs	Inquiry Parsing extracts line items from email/WhatsApp text into RFQ form
In	Suppliers	Smart Supplier Matching recommends suppliers based on category, history, and performance
In	WhatsApp	Incoming supplier quote images/PDFs processed by Quote OCR
In	Quotations	Quote Analysis scores and ranks supplier responses
In	Invoices / GRN	Document Classification auto-categorizes uploaded files
Out	RFQ Comparison	AI-extracted quotes feed directly into the comparison matrix
Out	Suppliers	Non-responsive supplier flagging triggers Auto Follow-Up
Out	Approvals	Anomaly Detection flags feed into the approval review process

Step-by-Step Guide

1. Inquiry Parsing

1. Navigate to RFQs > New RFQ.
2. Click Parse from Text (or paste into the AI input field).
3. Paste the raw text from a customer email, WhatsApp message, or any unstructured source.
4. Click Extract Items.
5. Tawridy's AI identifies and extracts: item descriptions, quantities, units, specifications, and any mentioned delivery requirements.
6. Review the extracted line items in the table. Edit or remove any incorrect entries.
7. Click Confirm to populate the RFQ form with the extracted data.

2. Quote OCR

1. Open an RFQ and navigate to the Quotes tab.
2. Click Upload Quote and select the supplier's quote document (PDF, image, or scanned file).
3. Click Extract with AI.
4. The AI OCR engine reads the document and extracts: item descriptions, unit prices, total prices, lead times, payment terms, and validity period.
5. Review the extracted data side-by-side with the original document.
6. Correct any misread values (common with handwritten or low-quality scans).
7. Click Save Quote. The extracted data is added to the RFQ comparison matrix.

3. Smart Supplier Matching

1. While creating or editing an RFQ, click Suggest Suppliers.
2. Tawridy's AI analyzes the RFQ item categories and recommends suppliers based on:
   • Category match (suppliers who deal in these item categories).
   • Historical performance (response rate, delivery %, quality score from the Supplier Scorecard).
   • Pricing competitiveness (historical pricing for similar items).
3. Each suggested supplier shows a match score and reasoning.
4. Select the suppliers you want to add to the RFQ and click Add.

4. Quote Analysis

1. After receiving quotes from multiple suppliers, open the RFQ Comparison Matrix.
2. Click AI Analysis.
3. The AI evaluates all quotes and provides:
   • A score for each supplier based on price, lead time, payment terms, and historical reliability.
   • A natural language recommendation summarizing which supplier offers the best overall value and why.
   • Flagged concerns (e.g., "Supplier X's price is 40% below market average for this item category").
4. Use the AI recommendation as input to your decision. Select the winning supplier manually.

5. Anomaly Detection

1. Anomaly Detection runs automatically in the background when quotes are entered.
2. The system flags:
   • Suspiciously low prices — prices significantly below market average or historical data, which may indicate counterfeit or substandard goods.
   • Abnormal lead times — delivery promises that are unrealistically short or excessively long compared to category norms.
   • Price spikes — sudden increases from a supplier compared to their own historical pricing.
3. Flagged anomalies appear as warning badges on the quote line item.
4. Click the badge to see the AI's explanation and supporting data.
5. Investigate and decide whether to proceed, request clarification from the supplier, or reject the quote.

6. Auto Follow-Up

1. Navigate to Admin > Settings > AI > Auto Follow-Up (or configure via WhatsApp/Email settings).
2. Enable the feature and set the timing rules:
   • First reminder: X hours after RFQ distribution.
   • Second reminder: Y hours after first reminder.
   • Maximum reminders: configurable (default 2).
3. The system runs a scheduled job (cron) that checks for non-responsive suppliers.
4. Reminders are sent via the supplier's preferred channel (WhatsApp or Email).
5. If a supplier responds, the follow-up chain stops automatically.

7. Document Classification

1. Upload any document to Tawridy (via RFQ, PO, GRN, or Supplier record).
2. The AI automatically classifies the document type:
   • Invoice — supplier or customer invoice.
   • Certificate of Conformance (CoC) — quality certification document.
   • Packing List — shipment contents listing.
   • Trade License — business registration document.
   • Quotation — supplier price quote.
   • Other — unrecognized document type.
3. The classification label appears on the document attachment.
4. You can override the classification if the AI categorized it incorrectly.

Key Fields Explained

Feature	Input	Output
Inquiry Parsing	Raw text (email, WhatsApp, pasted)	Structured line items (description, qty, unit, specs)
Quote OCR	PDF, image, or scanned document	Extracted prices, lead times, terms
Smart Supplier Matching	RFQ item categories	Ranked supplier suggestions with match scores
Quote Analysis	Multiple supplier quotes	Scored comparison with natural language recommendation
Anomaly Detection	Quote prices and lead times	Warning flags with explanations
Auto Follow-Up	Non-responsive supplier list	Automated reminder messages via WhatsApp/Email
Document Classification	Uploaded file (any format)	Document type label

Tips & Best Practices

TIP

Always review AI-extracted data before confirming. OCR accuracy is highest with clean, typed documents (95%+) and lower with handwritten or low-resolution scans. A quick review catches the occasional misread.

TIP

The more historical data Tawridy accumulates, the smarter the AI becomes. Supplier Matching and Anomaly Detection improve significantly after 3-6 months of usage as the system learns your pricing patterns and supplier behavior.

WARNING

AI recommendations are advisory, not authoritative. They are designed to support procurement professionals, not replace their judgment. Always apply your domain expertise, especially for high-value or safety-critical procurement decisions. Anomaly flags require human investigation, not automatic rejection.

• Use Inquiry Parsing to eliminate the copy-paste-format cycle when converting customer emails into RFQs. It saves 5-10 minutes per inquiry.
• Enable Auto Follow-Up with reasonable timing (24h first reminder, 48h second). It significantly improves supplier response rates without manual effort.
• When Quote OCR misreads a value, correct it and the system learns from the correction over time.
• Use Document Classification to keep your attachments organized. Well-classified documents make audit preparation much easier.

FAQ

Q: Does the AI send data outside of Tawridy? A: AI processing uses secure, privacy-compliant models. Document content is processed for extraction and classification only, never used for model training, and never shared with third parties.

Q: Can I disable specific AI features? A: Yes. Each AI feature can be toggled on or off independently in Admin > Settings > AI. Organizations that prefer fully manual workflows can disable all AI features.

Q: How accurate is the Quote OCR? A: For typed, well-formatted PDF quotes, accuracy typically exceeds 95%. For scanned or handwritten documents, accuracy varies between 80-90%. Always review extracted data before confirming.

Q: Does Smart Supplier Matching work for new organizations with no history? A: For new organizations, matching is based primarily on category alignment. As you accumulate RFQ, PO, and delivery data, the matching algorithm incorporates performance and pricing history for increasingly accurate recommendations.

Q: What languages does the AI support? A: AI features support English and Arabic text input. Inquiry Parsing and Quote OCR can process bilingual documents common in the GCC region.

Q: Will AI auto-reject a flagged anomaly? A: Never. Anomaly Detection raises warnings for human review. It does not reject quotes, block POs, or take any automated action. The decision always remains with your team.