Constraint Studio

Elevate plan

Constraint Studio is the rule editor behind PO hints and the constraint optimizer. You configure business rules in a guided workflow; the technical graph is still available through Advanced graph for inspection and power-user changes.

Constraint Studio with compact readiness chips for semantic roles — The readiness chips show which semantic roles are already mapped from Product Settings or Data Modeling.

What a rule controls

A constraint rule can:

Show line or PO-level hints when a PO is below MOQ, not in pack multiples, over budget, missing a group, or missing required data.
Act as an optimizer target, so the optimizer tries to close the gap by adding eligible variants or increasing existing lines.
Add an optimizer objective, so an already-valid plan is ranked toward a better fill or capacity outcome.

Rules are evaluated for Draft and Sent POs only. Other PO statuses keep their saved data, but active constraint rules do not apply there.

Constraint readiness

The readiness section checks whether common semantic roles are mapped:

Role	Used by
Group key	Group exact count, group minimum total, group coverage.
Unit CBM	PO max CBM, container fill, capacity blocks.
Unit weight	PO max weight and weight-based capacity rules.
Case size	Case-size and multiple-based planning.
Pallet quantity	Pallet or block-capacity planning.
Purchase family	Coverage, bundle, and family-based rules.
Product MOQ	Product-level MOQ summed across variants.

Mapped roles can come from Product & variant attributes or from Data Modeling semantic mappings. Hover a readiness chip to see the exact field source. Missing roles do not block saving every rule, but presets that need that role ask you to bind a field manually.

Presets

Click Add rule to open the preset picker.

The current presets are:

Preset	What it checks	Optimizer behavior
MOQ	Each line meets supplier-variant MOQ.	Can increase/add lines to meet MOQ.
Pack size / multiples	Each line is a valid pack multiple.	Final suggested quantities must respect multiples.
Min line quantity	Every line reaches a fixed minimum.	Can increase lines.
Max line quantity	Every line stays under a fixed maximum.	Fill gaps cannot reduce lines, so this appears as a warning/limit.
Supplier minimum value	PO value reaches a supplier minimum.	Can add/increase eligible lines.
Supplier minimum units	Total ordered units reach a supplier minimum.	Can add/increase eligible lines.
PO budget cap	PO value stays under a cap.	Fill gaps cannot reduce an over-budget PO.
PO max units	Total ordered units stay under a cap.	Hard limit for additive plans.
PO max CBM	Ordered units times unit CBM stay under a cap.	Hard limit when the rule is a target.
PO max weight	Ordered units times unit weight stay under a cap.	Hard limit when the rule is a target.
Container fill target	Total capacity measure reaches a target.	Adds a fill objective.
Exact group count	PO contains exactly the required number of groups.	Evaluates group count; additive fixes depend on candidate data.
Minimum units per variant	Every active variant reaches a shared minimum.	Can increase/add eligible variants.
Minimum total per group	Every active group reaches a summed minimum.	Can increase/add lines in underfilled groups.
Product MOQ	All variants of each product are summed and compared to a product-level MOQ field.	Can increase/add variants until product MOQ gaps are filled.
Group coverage	Active groups cover available supplier variants in the same group.	Can add missing variants from uncovered groups.
Ratio / bundle	A numerator measure divided by a denominator measure reaches a minimum ratio.	Evaluates the ratio; additive fixes depend on candidate data.
Fill capacity blocks	Current capacity block is underfilled and should move toward the next full block.	Objective-only by default.

Rule metadata

Every guided rule has the same high-level settings:

Setting	Behavior
Name	Merchant-facing rule name shown in settings, PO hints, and optimizer reasons.
Enabled	Disabled rules are saved but ignored.
Scope	General applies to all POs. Supplier scope applies only to selected active suppliers.
Statuses	Only Draft and Sent are valid active statuses. At least one must be selected.
Severity	Controls the PO hint tone: Info, Warning, or Error. It does not by itself make a rule hard.
Optimizer target	Makes the optimizer treat the rule as a target/limit. Warning-only rules still show hints but are not a target.

Field binding

Some presets can use built-in PO fields immediately. Others need a field binding:

Product MOQ needs a product identity field, ordered units, and a product-level MOQ field.
CBM and weight rules need a per-unit Number field.
Group rules need a group key or purchase family field.
Ratio rules need two numeric fields.

Constraint Studio suggests fields from semantic roles first. It does not silently bind an unsafe field when the role or type does not match; choose the field explicitly if the suggestion is missing.

Product MOQ preset with product identity, ordered units, and Product MOQ field bindings — One Product MOQ rule can use a product-level attribute value for every product instead of hard-coding thresholds per product.

PO max CBM preset with a Unit CBM field and maximum CBM threshold — Capacity rules multiply ordered quantity by a per-unit measure such as Unit CBM.

Preview and save

Before saving, preview a rule against sample data or a real PO. Real-PO preview is the best check because it uses your PO lines, supplier, fields, and current saved values.

Preview and evaluation do not change PO quantities. They return pass/fail state, severity, and missing-data reasons such as missing Product MOQ, Unit CBM, Unit weight, Group key, or Purchase family.

Settings saves are versioned. If another tab changes the constraint settings first, Logistified rejects the stale save and asks you to reload instead of overwriting someone else’s rules.

Advanced graph nodes

Guided presets compile to a technical graph. Most teams never need to edit it, but it is useful for debugging.

Datasets

Dataset	What it contains
`line_items`	Current PO lines after data-quality exclusions. Use this for normal rule checks.
`all_line_items`	Current PO lines before exclusions. Use this when the rule itself must report missing data.
`supplier_variants`	Eligible supplier catalog variants, including variants not yet on the PO.
`combined_line_items`	Current PO lines plus simulated optimizer additions.
`candidate_line_items`	Only simulated optimizer candidate rows.

Supported nodes

Node	Input -> output	Main settings
`meta.rule`	None -> metadata	Internal rule metadata; not needed in guided rules.
`source.po`	None -> object	Current PO context: header fields, current lines, supplier variants, optimizer candidates. A normal graph has one.
`source.dataset`	None -> dataset	Reads a named dataset. This is valid for legacy/advanced graphs but hidden from the normal library.
`source.field`	None -> scalar	Reads one field path, for example `line_item.quantityOrdered` or a mapped Product Settings/Data Model field.
`source.value`	None -> scalar	Static text, number, boolean, or date value.
`transform.extract`	PO object/dataset -> dataset or scalar	Selects `line_items`, `all_line_items`, `supplier_variants`, `combined_line_items`, `candidate_line_items`, or a nested PO field.
`transform.filter`	Dataset + optional compare input -> dataset	Filters rows by field path, operator, and right value.
`transform.map`	Dataset -> dataset	Adds or replaces a row value. Modes: copy `field`, write `constant`, or calculate `math` into an output field.
`transform.group_by`	Dataset -> grouped dataset	Groups rows by a field such as product ID, variant ID, group key, or purchase family.
`transform.aggregate`	Dataset/grouped dataset -> scalar or grouped values	`sum`, `count`, `distinct_count`, `min`, `max`, `avg`; grouped input returns one result per group.
`transform.set_operation`	Dataset -> scalar/set/boolean	`distinct_values`, `distinct_count`, `cardinality_in_set`, `contains_all`, `contains_any`.
`logic.math`	Scalars -> scalar	`+`, `-`, `*`, `/`, `min`, `max`, `ceil`, `floor`, `round`, `abs`. Used for ratios and capacity blocks.
`logic.compare`	One or two values -> boolean	`>=`, `>`, `<=`, `<`, `==`, `!=`, `contains`, `not_contains`, `in_set`, `before`, `after`; pack-size presets also generate `multiple_of`.
`logic.group_compare`	Grouped left + grouped/right limits -> boolean	Compares each group to its own limit. Missing-right behavior can fail, ignore, or use a default. Product MOQ uses this node.
`logic.boolean`	Booleans -> boolean	`and`, `or`, `not`, `xor`.
`logic.date`	Dates -> scalar or boolean	`before`, `after`, `diff_days`.
`logic.quantifier`	Dataset -> boolean	Requires `all`, `any`, or `none` of the rows to satisfy a row comparison. Line MOQ and min/max line quantity use this.
`logic.coverage`	Current rows + universe rows -> boolean	Checks whether PO groups/variants cover the available supplier universe. Can check only groups already present or the full universe.
`output.finding`	Boolean -> PO hint/badge	Emits pass/fail text, severity, affected fields, affected variants/groups, and missing-data details. It never changes PO data.
`output.optimizer_score`	Scalar -> optimizer objective	Adds a numeric ranking signal for candidate plans. Hard feasibility still wins over score.

Malformed graphs, cycles, missing nodes, wrong input/output types, missing source.po, and finding outputs that are not driven by logic nodes are rejected. Use the guided workflow unless you need a graph-level rule.

Where rules appear on POs

Saved rules appear in three places:

The Constraints column on PO line items.
Purchase-order hints in the guided PO preview area.
The Optimize dialog, where target rules, remaining reasons, and proposed actions are explained.