Inventory Optimizer Agent Skill

Inventory Optimizer

Core Principles

Inventory is cash sitting on shelves. Too much ties up capital and spoils. Too little loses sales and frustrates customers. The Inventory Optimizer skill treats stock as a dynamic system — balancing par levels, usage velocity, supplier lead times, and seasonal demand fluctuations.

The Five Pillars of Smart Inventory

• Know your par levels. Every item has a minimum quantity you should never dip below. Set pars based on historic usage and reorder lead time.
• Track usage velocity. Fast-movers need frequent reordering. Slow-movers need scrutiny. Group items by turnover rate (A = high, B = medium, C = low).
• Account for lead time variance. A supplier who delivers in 2 days on paper often takes 5. Build buffer based on actual delivery history.
• Seasonal forecasting. Demand shifts with holidays, weather, and local events. Adjust pars proactively, not reactively.
• First-expiry-first-out (FEFO). Rotate stock religiously. The oldest inventory moves first. Nothing wrecks margins like spoilage.

Skill Workflow

Step 1 — Audit Current Inventory

Gather the current state by asking the user or pulling from their POS/ERP system. Collect:

• Current stock quantities for each item
• Unit of measurement (kg, cases, bottles, units)
• Storage location (dry, cold, frozen)
• Expiry dates (for perishables)
• Last 30/60/90 days of usage data
• Current supplier and price per unit

Prompt the user for:

> "Let's start with your inventory snapshot. Do you have a stock count sheet, a POS report, or should I help you build one from scratch?"

Step 2 — Calculate Par Levels & Reorder Points

For each item, calculate:

Reorder Point = (Average Daily Usage × Lead Time in Days) + Safety Stock
Safety Stock = Average Daily Usage × Lead Time Variance Buffer
Par Level = Reorder Point + (Average Daily Usage × Order Cycle Days)

Example calculation for a busy café:

• Espresso beans: 3kg/day usage, 4-day lead time, 2-day variance buffer
• Reorder Point = (3 × 4) + (3 × 2) = 18kg
• Order cycle: weekly (7 days)
• Par Level = 18 + (3 × 7) = 39kg
• If current stock = 12kg → trigger order for 27kg (Par − Current)

Step 3 — Classify Items (ABC Analysis)

| Class | % of Items | % of Cost | Strategy |

|-------|-----------|-----------|----------|

| A | 10-20% | 70-80% | Daily tracking, tight pars, frequent small orders |

| B | 20-30% | 15-20% | Weekly review, standard pars |

| C | 50-70% | 5-10% | Monthly review, bulk order, low scrutiny |

Action: Generate the ABC classification table and recommend ordering frequency by class.

Step 4 — Identify Slow-Movers & Waste

Cross-reference usage data with expiry dates. Flag items where:

• Usage rate < 1 unit per week (slow-mover — consider delisting)
• > 20% of stock expires within 7 days (waste risk — promote or discount)
• Usage dropped > 30% vs. prior period (demand shift — reduce order)

Recommend actions:

• Promote: Feature on specials board, bundle with fast-movers
• Reduce: Cut order quantity by 50% for next cycle
• Delist: Stop ordering and sell through remaining stock

Step 5 — Generate Purchase Order

Compile all items at or below reorder point into a structured PO:

| Item | Current Stock | Par Level | Order Qty | Supplier | Unit Price | Total |

|------|-------------|-----------|-----------|----------|-----------|-------|

| Espresso Beans | 12kg | 39kg | 27kg | Bean Co. | $14/kg | $378 |

| Milk (2%) | 8 gal | 22 gal | 14 gal | Dairy Fresh | $3.50/gal | $49 |

Total order value: $427

Ask the user:

> "Here's your draft order. I've calculated quantities based on par levels and current usage. Would you like to adjust any items before I finalize?"

Step 6 — Schedule Recurring Reviews

Set reminders based on item class:

• A items: Review daily or every other day
• B items: Weekly review
• C items: Monthly review

Offer to schedule:

> "I can set up a daily stock check reminder for your A-items and a weekly full inventory review. Would you like me to schedule that?"

Trigger Phrases

| Phrase | What Happens |

|--------|-------------|

| "Check my inventory levels" | Runs full inventory audit against pars |

| "What should I reorder?" | Generates purchase order for items below reorder point |

| "I'm running low on [item]" | Checks current stock vs par for that item, recommends order qty |

| "Show me slow-moving items" | Runs ABC analysis, highlights C-items and waste risks |

| "Do an inventory review" | Full workflow: audit → classify → flag risks → generate PO |

| "Set par levels for [category]" | Guides user through setting pars for a specific category |

Key Instructions Summary

• Audit: Get current quantities, usage data, and supplier info
• Calculate: Apply reorder point and par level formulas
• Classify: ABC analysis to prioritize attention
• Flag: Identify slow-movers, waste risk, and demand shifts
• Generate: Produce a formatted purchase order with totals
• Schedule: Set recurring review reminders by item class
• Iterate: After each order cycle, compare projected vs actual usage and adjust pars

Common Mistakes

• Ordering on intuition. "It feels like we're low" leads to overstock or shortages. Always use data.
• Ignoring lead time variance. A supplier who's late 20% of the time needs a bigger buffer.
• One-size-fits-all pars. A café uses more milk in summer (iced drinks). Adjust pars seasonally.
• Not tracking waste. If you don't measure what's thrown away, you can't fix it.
• Skipping the ABC analysis. Spending equal time on napkins and prime rib is inefficient.

*"Inventory is money sitting on shelves — make every dollar work."*