Intelligent Packing API

The Intelligent Packing API provides AI-enhanced 3D bin packing built on a three-layer hybrid architecture. It combines a deterministic rule engine, an AI matching/pre-screening layer, and an OR-Tools mathematical solver.

Overview

The service processes each request through three layers:

Layer 1 — Rule Engine: handles pre-assigned pallets and filters out unavailable / unfit pallet types.
Layer 2 — AI Model: SKU–pallet matching, pallet pre-screening, and solution recommendations.
Layer 3 — OR-Tools: mathematically guaranteed bin-packing solve.

Additional capabilities:

AI pallet pre-screening (reduce a large pallet catalog to an optimal subset)
Parallel and batch processing for large item counts
Dynamic entropy tuning
LDM (Loading Meter) output for logistics billing

Base URL:

https://api.alaikis.com

All endpoints are mounted under both /api/intelligent-packing and /api/v1/intelligent-packing (v1 alias), and require authentication.

Endpoints

Method	Path	Description
POST	`/api/intelligent-packing/pack`	Run intelligent packing (three-layer solve)
POST	`/api/intelligent-packing/pack/preview`	Preview a plan without solving
POST	`/api/intelligent-packing/filter-pallets`	Filter unavailable / unfit pallet types
POST	`/api/intelligent-packing/prescreen-pallets`	AI pallet pre-screening (top-K subset)
GET	`/api/intelligent-packing/health`	Health check

Pack

Run the full three-layer intelligent packing solve.

Endpoint:

POST /api/intelligent-packing/pack

Request Body:

{
  "skus": [
    {
      "id": "SKU001",
      "weight": 5.5,
      "weight_unit": "kg",
      "length": 50,
      "width": 30,
      "height": 20,
      "length_unit": "cm",
      "quantity": 10,
      "is_dangerous": false,
      "is_stackable": true
    }
  ],
  "pallets": [
    {
      "id": "PALLET_A",
      "length": 120,
      "width": 100,
      "height": 150,
      "weight": 20,
      "max_weight": 500,
      "length_unit": "cm",
      "weight_unit": "kg",
      "cost": 5.0
    }
  ],
  "use_ai": true,
  "entropy_factor": 1.0,
  "time_limit_ms": 60000,
  "ldm_width": 2.44,
  "ldm_mode": "projection"
}

SKU Fields (`skus[]`)

Field	Type	Required	Default	Description
`id`	string	Yes	—	SKU identifier
`weight`	float	Yes	—	Item weight
`weight_unit`	string	No	`"kg"`	Weight unit
`length` / `width` / `height`	float	Yes	—	Item dimensions
`length_unit`	string	No	`"cm"`	Dimension unit
`quantity`	integer	No	`1`	Number of items
`is_dangerous`	boolean	No	`false`	Dangerous goods flag (isolated during packing)
`is_stackable`	boolean	No	`true`	Whether the item can be stacked

Pallet Fields (`pallets[]`)

Field	Type	Required	Default	Description
`id`	string	Yes	—	Pallet type identifier
`length` / `width` / `height`	float	Yes	—	Pallet dimensions
`weight`	float	No	`0.0`	Pallet tare weight
`max_weight`	float	Yes	—	Maximum load weight
`length_unit`	string	No	`"cm"`	Dimension unit
`weight_unit`	string	No	`"kg"`	Weight unit
`cost`	float	No	`0.0`	Pallet cost

Request Options

Parameter	Type	Default	Range	Description
`use_ai`	boolean	`true`	—	Enable AI enhancement
`entropy_factor`	float	`1.0`	0.8–2.0	Packing entropy factor
`time_limit_ms`	integer	`60000`	1000–300000	Solver time limit (ms)
`enable_parallel`	boolean	`true`	—	Enable parallel processing
`prescreen_pallets`	boolean	`true`	—	Enable AI pallet pre-screening
`top_k_pallets`	integer	`15`	5–100	Pallet subset size after pre-screening
`max_workers`	integer	`8`	1–16	Max parallel worker threads
`pre_assigned_pallets`	array	`null`	—	Pre-assigned pallets (see below)
`unavailable_pallet_types`	string[]	`null`	—	Pallet type IDs to exclude
`ldm_width`	float	`null`	> 0	LDM normalization vehicle width (m); defaults to `2.4`
`ldm_mode`	string	`"projection"`	—	LDM mode: `"projection"` or `"arrangement"`

::: tip Pre-assigned pallets pre_assigned_pallets[] items use the shape: \{ "pallet_type_id": "PALLET_A", "pallet_id": "P-1", "items": [\{ "sku_id": "SKU001", "quantity": 4 }], "used_volume": 0.0, "used_weight": 0.0 }. :::

Response:

{
  "pallets": [
    {
      "id": "pallet_1",
      "pallet_id": "pallet_1",
      "pallet_type": "PALLET_A",
      "items": [
        { "sku_id": "SKU001", "quantity": 10 }
      ],
      "total_weight": 75.0,
      "total_volume": 0.347,
      "utilization": 19.3,
      "source": "computed",
      "ldm": 0.492
    }
  ],
  "total_pallets": 1,
  "total_weight": 75.0,
  "total_volume": 0.347,
  "processing_time_ms": 12.3,
  "layers_used": ["L1_RULE_ENGINE", "L2_AI_MODEL", "L3_OR_TOOLS"],
  "ai_recommendations": null,
  "stats": {
    "total_pallets": 1,
    "total_weight": 75.0,
    "total_volume": 0.347,
    "total_ldm": 0.492,
    "ldm_mode": "projection",
    "ldm_width_used": 2.44
  },
  "prescreen_stats": null,
  "performance_stats": {},
  "errors": [],
  "warnings": [],
  "success": true,
  "message": "装托成功",
  "total_ldm": 0.492,
  "ldm_mode": "projection",
  "ldm_width_used": 2.44
}

Response Fields

Field	Type	Description
`pallets[]`	array	Solved pallets
`pallets[].id` / `pallet_id`	string	Pallet instance ID (kept in sync)
`pallets[].pallet_type`	string	Source pallet type ID
`pallets[].items`	array	`\{ sku_id, quantity }` placed on the pallet
`pallets[].total_weight`	float	Loaded weight (kg)
`pallets[].total_volume`	float	Loaded volume (m³)
`pallets[].utilization`	float	Volume utilization (%)
`pallets[].source`	string	`"computed"` or `"pre_assigned"`
`pallets[].ldm`	float	Per-pallet loading meter
`total_pallets`	integer	Number of pallets used
`layers_used`	string[]	Processing layers applied
`stats`	object	Aggregated statistics (weights, volumes, LDM, etc.)
`total_ldm`	float	Total loading meters for the solution
`ldm_mode`	string	LDM mode used
`ldm_width_used`	float	Actual vehicle width used (m)
`success`	boolean	Whether the solve succeeded
`message`	string	Human-readable status

LDM (Loading Meter)

LDM (Loading Meter) measures the equivalent floor length a solution occupies in a vehicle — the standard billing metric for European road freight.

Request Parameters (top-level, alongside skus / pallets)

Parameter	Type	Default	Description
`ldm_width`	float	`null`	Vehicle width in meters. When omitted, defaults to `2.4`.
`ldm_mode`	string	`"projection"`	Calculation mode: `"projection"` or `"arrangement"`

::: warning Vehicle width on this endpoint The /pack request has no vehicle field. Vehicle width therefore resolves as ldm_width (if provided) otherwise the 2.4 m default. Pass ldm_width explicitly to override. :::

Response Fields

Field	Type	Description
`total_ldm`	float	Total loading meters for the solution
`ldm_mode`	string	Mode used for calculation
`ldm_width_used`	float	Actual vehicle width used (meters)
`pallets[].ldm`	float	Per-pallet loading meter

projection Mode

ldm = (pallet_length_m * pallet_width_m) / ldm_width_m
total_ldm = Σ ldm

Example: a 1.2 m × 1.0 m pallet with ldm_width = 2.44 yields ldm = 1.2 * 1.0 / 2.44 = 0.492.

arrangement Mode

per_row   = max(1, floor(ldm_width_m / pallet_width_m))
rows      = ceil(total_pallets / per_row)
total_ldm = rows * pallet_length_m / ldm_width_m

arrangement assumes pallets are placed side-by-side along the vehicle width.

Example Request with LDM

payload = {
    "skus": [
        {"id": "SKU001", "weight": 5.5, "length": 50, "width": 30, "height": 20, "quantity": 10}
    ],
    "pallets": [
        {"id": "PALLET_A", "length": 120, "width": 100, "height": 150, "max_weight": 500}
    ],
    "ldm_width": 2.44,
    "ldm_mode": "projection"
}

Preview

Preview a packing plan without running the full solve. Returns valid SKU / available pallet counts, AI recommendations, and an estimated pallet count.

Endpoint:

POST /api/intelligent-packing/pack/preview

The request body is the same shape as /pack.

Response:

{
  "valid_skus": 1,
  "available_pallets": 1,
  "pre_assigned_pallets": 0,
  "ai_recommendations": null,
  "estimated_pallets": 2,
  "warnings": [],
  "preview": true
}

Filter Pallets

Filter out pallet types that are marked unavailable or that no SKU can fit into.

Endpoint:

POST /api/intelligent-packing/filter-pallets

Request Body:

{
  "skus": [
    {"id": "SKU001", "weight": 5.5, "length": 50, "width": 30, "height": 20, "quantity": 10}
  ],
  "pallets": [
    {"id": "PALLET_A", "length": 120, "width": 100, "height": 150, "max_weight": 500}
  ],
  "unavailable_pallet_types": ["PALLET_X"]
}

Response:

{
  "available_pallets": [
    {"id": "PALLET_A", "length": 120, "width": 100, "height": 150, "max_weight": 500}
  ],
  "filtered_count": 0,
  "total_count": 1,
  "filter_reasons": []
}

filter_reasons[] entries carry a pallet_id and a reason of "marked_unavailable" or "no_sku_fits".

Prescreen Pallets

AI pallet pre-screening — reduce a large pallet catalog to an optimal top-K subset before solving.

Endpoint:

POST /api/intelligent-packing/prescreen-pallets?top_k=20&use_ai=true

Query Parameter	Type	Default	Description
`top_k`	integer	`20`	Number of pallet types to keep
`use_ai`	boolean	`true`	Enable AI scoring

The request body uses the same skus / pallets / unavailable_pallet_types shape as Filter Pallets.

Response:

{
  "original_pallets": [ /* PalletType[] */ ],
  "filtered_pallets": [ /* PalletType[] */ ],
  "prescreen_stats": {},
  "performance_stats": {}
}

Health

Endpoint:

GET /api/intelligent-packing/health

Response:

{
  "status": "healthy",
  "service": "intelligent-packing",
  "version": "2.1.0",
  "layers": ["L1_RULE_ENGINE", "L2_AI_MODEL", "L3_OR_TOOLS"],
  "features": ["ai_prescreen", "parallel_processing", "performance_stats"]
}

Example Workflow

import requests

headers = {"Authorization": "Bearer <your-jwt-token>"}
base = "https://api.alaikis.com/api/intelligent-packing"

skus = [
    {"id": "SKU001", "weight": 5.5, "length": 50, "width": 30, "height": 20, "quantity": 10}
]
pallets = [
    {"id": "PALLET_A", "length": 120, "width": 100, "height": 150, "max_weight": 500}
]

# 1) Preview to estimate pallet count
preview = requests.post(f"{base}/pack/preview",
                        json={"skus": skus, "pallets": pallets},
                        headers=headers).json()
print("Estimated pallets:", preview["estimated_pallets"])

# 2) Run the full solve with LDM
result = requests.post(f"{base}/pack",
                       json={
                           "skus": skus,
                           "pallets": pallets,
                           "ldm_width": 2.44,
                           "ldm_mode": "projection"
                       },
                       headers=headers).json()

print("Pallets used:", result["total_pallets"])
print("Total LDM:", result["total_ldm"])
for pallet in result["pallets"]:
    print(pallet["pallet_type"], "->", pallet["utilization"], "% | ldm", pallet["ldm"])

Performance Notes

Large item counts automatically switch to batch mode; medium counts use parallel processing.
Pre-screening (prescreen_pallets) is applied automatically when the pallet catalog exceeds top_k_pallets.
Use time_limit_ms to cap solver time for real-time scenarios.

Overview​

Endpoints​

Pack​

SKU Fields (skus[])​

Pallet Fields (pallets[])​

Request Options​

Response Fields​

LDM (Loading Meter)​

Preview​

Filter Pallets​

Prescreen Pallets​

Health​

Example Workflow​

Performance Notes​

Overview

Endpoints

Pack

SKU Fields (`skus[]`)

Pallet Fields (`pallets[]`)

Request Options

Response Fields

LDM (Loading Meter)

Preview

Filter Pallets

Prescreen Pallets

Health

Example Workflow

Performance Notes