Skip to main content

See the AI in action

Explore how LumberFlow’s AI agent handles your lumber procurement workflow end-to-end.

​
Intelligent Email Processing

LumberFlow’s AI agent monitors your connected Gmail inbox in real time, processes every supplier email automatically, and extracts structured data — but you always have the final say before anything enters your pipeline.

​
Real-Time Gmail Integration

LumberFlow uses Gmail push notifications (webhooks) rather than polling, which means:
  • Zero lag: Supplier emails are detected and processed the moment they arrive
  • No manual forwarding: Your connected Gmail inbox is monitored automatically
  • Background processing: Quote extraction happens while you’re doing other things
  • Fallback safety: A scheduled cron job provides backup processing every 15 minutes if the webhook isn’t triggered
To enable real-time processing, connect your Gmail account under Settings → Integrations.
The AI handles every format suppliers use:
  • Email body text: Price lists formatted as tables, bullet points, or free-form paragraphs
  • PDF attachments: Structured quote sheets and price lists
  • Excel/spreadsheet attachments: Column-based pricing tables
  • HTML emails: Formatted supplier quote templates
For each format, the AI extracts: dimensions, species, grade, quantities by length, unit price ($/MBF), total price, lead time, payment terms, and validity date.
The AI is trained on lumber industry terminology:
  • Standard abbreviations: SPF, SYP, DF, HF, KD, S4S, S1S2E
  • Grade designations: Construction, Standard, Utility, Stud, Select Structural, No.1&Btr, Clear
  • Dimension and length formats: 2x4, 2x6, 2x8, 2x10, 2x12 × 8ft/10ft/12ft/16ft/20ft
  • Pricing formats: per piece, per MBF (thousand board feet), per pack — all normalized to $/MBF
  • Delivery terms: FOB mill, FOB destination, freight included/excluded

​
Human-in-the-Loop Validation

Every AI extraction goes through a human validation step before entering your pipeline. This is not optional — it’s a core design principle.
1

AI Extraction

System processes the supplier email or attachment and extracts all structured data
2

Validation Interface

A two-panel interface appears: the original supplier document on the left, the AI-extracted data on the right
3

Human Review

You compare the extraction against the source document and correct any errors
4

Approval & Integration

Approved data is saved to the quote, the sourcing card updates, and the pipeline advances

​
Validation Best Practices

Product Specifications: Always verify that AI correctly identified dimensions, species, and grades — these determine whether a quote is actually for the right product
Pricing Accuracy: Confirm unit prices are in /MBF,not/MBF, not /pack or /piece.LumberFlowalwaysstorespricesin/piece. LumberFlow always stores prices in /MBF for consistent comparison.
Delivery Terms: Review extracted lead times and FOB terms carefully — these affect total landed cost and should factor into your purchase decision
Never skip validation: Even when AI accuracy is high, the cost of an error in a lumber purchase order is significant. Validate every quote before committing.

​
AI-Generated Communications

​
RFQ Email Generation

When you send an RFQ from the Kanban board, the AI generates professional, supplier-specific emails from your procurement plan specifications:

Supplier-Specific Drafts

Each email is drafted for the specific supplier, referencing your relationship history and their specialty areas

Complete Specifications

Full product requirements (dimension, species, grade, lengths, quantities) and your target pricing

Professional Tone

Industry-appropriate language that maintains your professional image with every supplier

Review Before Send

All AI-generated emails are shown for review before sending — you can edit any draft
RFQ generation uses parallel AI processing (up to 5 concurrent drafts) with per-user rate limiting and deduplication. If you accidentally click “Send” twice, the system prevents duplicate emails to suppliers.

​
Counter-Offer Drafting

When you draft a counter-offer, the AI:
  • Reads the full email thread: Summarizes prior communications between you and the supplier
  • References previous offers: The counter-offer acknowledges the supplier’s last quote without you having to write the context
  • Maintains reply threading: Sends as a proper email reply (with RFC Message-ID threading), so it appears in the right email thread in the supplier’s inbox
  • Adapts tone: Adjusts formality based on your relationship with the supplier
You control the negotiation strategy before the AI drafts:
  • Target price: Set your desired price and acceptable range
  • Custom instructions: Add specific requirements, volume commitments, or timing constraints
  • Tone selection: Professional, friendly, or direct communication styles
  • Full edit capability: Review and edit the complete draft before sending

​
Negotiation Templates

Initial Counter-Offer

First-time price negotiations

Follow-Up Negotiation

Ongoing discussion rounds with thread context

Volume Discounts

Quantity-based pricing requests

Delivery Terms

Shipping and timeline negotiations

Payment Terms

Extended payment discussions

Final Acceptance

Agreement confirmation language

​
Market Sentiment Intelligence

The Market Sentiment Nudge surfaces species-level market intelligence directly on your Kanban cards and in your pre-purchase-order flow — so every sourcing decision is informed by current market conditions.
For each species you’re sourcing, the AI analyzes:
  • Lumber news and trade publications: Mill openings/closures, supply chain events, trade policy changes
  • Mill capacity signals: Production disruptions, capacity additions, inventory levels
  • Demand indicators: Housing starts, building permits, construction activity
  • Price trend data: Recent price movements and momentum
This produces a species-level sentiment score (Bullish / Neutral / Bearish) with a confidence level and a daily summary of the signals driving the assessment.
Every sourcing card displays a sentiment badge for its species:
  • Bullish 🟢: Market conditions favor suppliers — consider buying sooner, before prices rise
  • Neutral ⚪: No strong directional signal — standard procurement timing applies
  • Bearish 🔴: Market conditions favor buyers — you may have pricing leverage, or conditions may improve
Click See why on any badge to read the AI’s supporting analysis and the news events driving the signal.
Disagree with the AI’s sentiment assessment? Submit feedback directly from the card. Your feedback helps calibrate the model over time.

Learn more about lumber market analysis

See how LumberFlow delivers daily market intelligence for lumber buyers.

​
Team Insights: Your Internal Price Benchmark

Team Insights uses your organization’s own historical quote data as pricing benchmarks — normalized to $/MBF for consistent comparison across SKUs.
In the sourcing request drawer and on the Team Insights dashboard:
  • Median price your team has paid for a given SKU (dimension-species-grade group) in the last 90 days
  • In-flight median: What open (not-yet-fulfilled) quotes are currently priced at — useful for understanding where active negotiations are landing
  • Grade filters: Compare benchmarks across different grade levels for the same species and dimension
  • Supplier breakdown: Which suppliers are typically competitive for a given SKU
  • All prices are normalized to $/MBF using the getMBFPerPack() utility — so a 2x4 and a 2x6 quote are always comparable on a per-board-foot basis
  • Business-day windows: Benchmarks use a 2-business-day rolling window to avoid stale data
  • SKU group normalization: Dimension-species-grade combinations are normalized so “2x4 SPF Stud” and “2x4 SPF Std&Btr” are tracked separately
  • Organization-scoped: You only see your organization’s data — never other customers’ data
Team Insights benchmarks are recomputed every 30 minutes via a background batch job. New quotes you validate and approve are reflected in the next benchmark cycle.

​
Email Thread Intelligence

LumberFlow uses RFC Message-ID threading to maintain proper email continuity:
  • Every RFQ email is sent with a unique Message-ID header
  • Counter-offers are sent as replies with proper In-Reply-To and References headers
  • Supplier responses are matched back to the original RFQ thread automatically
  • The full thread history is preserved for counter-offer context generation
When you send bulk RFQs or counter-offers:
  • Each supplier receives their own independent email thread
  • Bulk operations never cross-contaminate threads — SPF Mill A never sees SPF Mill B’s pricing
  • Counter-offer context is generated separately per supplier, using that supplier’s conversation history
  • Response rates and timing are tracked per supplier for performance analytics

​
Quality Assurance

95%+ Accuracy

Pricing and specification extraction accuracy across standard supplier email formats

100% Validation

All AI-extracted data requires human approval before entering your pipeline

Rate Limited

Per-user AI rate limiting prevents runaway costs (10 RFQ generations/min, 20 quote processings/min)

Idempotent Operations

Duplicate-prevention at every AI step — no double-sent RFQs, no double-processed quotes

​
Troubleshooting AI Features

Problem: AI misreads pricing or specificationsSolutions:
  • Use the validation interface to correct errors before approving
  • If a supplier’s format is consistently misread, contact support — we can tune extraction for specific formats
  • Ask suppliers to use consistent formatting (e.g., always specify prices as $/MBF explicitly)
Problem: Supplier emails not appearing in the board quicklySolutions:
  • Verify Gmail is connected under Settings → Integrations
  • Check that the supplier’s reply is sent from an email address registered in your supplier contact list
  • The fallback cron job runs every 15 minutes — if the webhook missed a message, it will be caught in the next cycle
  • Contact support if emails are still missing after 30 minutes
Problem: Generated counter-offer emails need significant editingSolutions:
  • Add specific instructions in the custom instructions field (e.g., “Reference our Q1 volume commitment of 500 packs”)
  • Update your organization profile with your preferred communication style
  • For recurring scenarios, save effective counter-offer language and paste it into the custom instructions field

​
Next Steps

Dashboard Overview

See how AI insights — sentiment, benchmarks, and KPIs — appear on your Kanban board

Advanced Workflows

Master quote validation, counter-offer negotiation, and purchase order creation

Best Practices

Optimize your AI-assisted procurement workflow for maximum efficiency

Troubleshooting

Resolve common AI processing issues