Perimattic

Demand OS Features

Every feature of our demand planning artificial intelligence software, explained

Sixteen capabilities across four disciplines. Every feature of Demand OS maps to a real failure mode in enterprise demand planning — and because it is artificial intelligence software that shows its work, every output can be verified, overridden, and audited.

Book a Demo
Demand planning analytics dashboard

What features does Perimattic Demand OS include?

Perimattic Demand OS includes sixteen capabilities across four disciplines. The AI Forecasting Engine auto-selects the best statistical or ML model per SKU, while Demand Sensing sharpens the near-term forecast with live POS and external signals. Scenario Planning lets teams model optimistic, pessimistic, and most-likely outcomes before committing to a number. Inventory Optimization converts forecast confidence into service-level-driven safety stock recommendations that balance working capital against stockout risk. Every output is backed by Explainable AI — a ranked, plain-language list of the factors behind each forecast — and the AI Copilot gives planners a natural-language interface to query, override, and audit the plan.

JUMP TOForecasting & SensingPlanning & CollaborationInventory & SupplyGovernance & Trust
01

Forecasting & Sensing

The engine layer of the demand planning artificial intelligence software: an ensemble of statistical and ML models turning demand history and live signals into a per-SKU forecast you can defend.

Forecasting & Sensing
ML Forecasting
Ensemble of statistical and machine-learning models, auto-selected per SKU based on data patterns and forecast horizon. No single model is trusted blindly; the winner is logged with its reason.
Solves: One-size-fits-all models that fit fast movers and fail long-tail SKUs.
Read the deep dive →
Demand Sensing
Short-horizon signal detection using POS, order, and external data to sharpen the near-term forecast before the week closes.
Solves: Weekly planning cycles that miss demand spikes until it is too late to react.
Read the deep dive →
New Product Forecasting
Cold-start forecasting for items with zero or sparse history, using analogous product matching and Bayesian priors. Planners see which analogues were used and why.
Solves: Launch forecasts built on gut feel because there is no sales history.
Read the deep dive →
ABC-XYZ Segmentation
Automatic segmentation by value and demand variability, so model selection, safety stock, and review cadence all match the SKU profile.
Solves: Planners spending equal time on A-items and C-items alike.
Read the deep dive →
02

Planning & Collaboration

The workflow layer: AI-generated scenarios and structured consensus produce one number the whole business can stand behind, with full traceability.

Planning & Collaboration
Scenario Planning
Model optimistic, pessimistic, and most-likely demand scenarios side-by-side. Commit to a plan only when you have seen the range.
Solves: Single-point forecasts that hide the risk in the plan.
Read the deep dive →
What-If Analysis
Adjust any driver — promotional uplift, price change, channel mix — and see the downstream inventory and supply impact instantly.
Solves: Waiting a full cycle to learn what a planning decision costs.
Read the deep dive →
Consensus Forecasting
Sales, Finance, and Ops submit inputs into a single structured workflow. One number, full traceability, no more spreadsheet reconciliation.
Solves: Consensus meetings that reconcile data instead of making decisions.
Read the deep dive →
Promotion Planning
Model promotional uplift, cannibalization across SKUs, and halo effects, so the promotional forecast is built on data, not gut feel.
Solves: Flat percentage uplifts applied to every promotion.
Read the deep dive →
03

Inventory & Supply

The execution layer: the software converts AI forecast confidence into working-capital decisions — safety stock, supply plans, and replenishment orders.

Inventory & Supply
Inventory Optimization
Service-level–driven safety stock recommendations that balance working capital against stockout risk, by SKU, location, and season.
Solves: Blanket safety-stock rules that overstock some SKUs and starve others.
Read the deep dive →
Supply Planning
Converts optimized demand plans into constrained supply requirements, accounting for lead times, MOQs, and supplier capacity.
Solves: Demand plans that ignore what suppliers can actually deliver.
Read the deep dive →
Automated Replenishment
Rules-based and ML-driven replenishment triggers that reduce manual PO creation while maintaining agreed service levels.
Solves: Planners hand-keying hundreds of routine purchase orders.
Read the deep dive →
Demand Shaping
Recommends pricing, promotion, or channel actions to shift demand toward available supply, turning a constraint into a commercial lever.
Solves: Treating supply constraints as pure loss instead of a steerable outcome.
Read the deep dive →
04

Governance & Trust

The trust layer: explainable artificial intelligence, bias detection, and full audit trails — the reason a CFO, an auditor, and a skeptical planner can all sign off.

Governance & Trust
Explainable AI
Every forecast includes a ranked list of contributing factors — seasonality, trend, promotions, external signals — in plain language, not just a number.
Solves: Black-box outputs planners cannot verify or confidently override.
Read the deep dive →
Forecast Accuracy Dashboard
Track MAPE, WMAPE, bias, and Forecast Value Add at every level — SKU, category, channel, region — with drill-down to the root cause.
Solves: Accuracy reporting that says what happened but never why.
Read the deep dive →
Bias Detection
Automatically surfaces systematic over- or under-forecasting by planner, product group, or channel so teams correct it before it compounds.
Solves: Persistent optimism or sandbagging hiding inside aggregate accuracy.
Read the deep dive →
Audit Logs & Versioning
Every override, approval, and plan change is logged with a timestamp and reason. Full version history for compliance, review, and learning.
Solves: final_v3_FINAL.xlsx and untraceable changes to the number.
Read the deep dive →

Why It Matters

What demand planning artificial intelligence software changes in practice

Features are means, not ends. Across the sixteen capabilities above, the outcomes teams measure are the same three numbers.

20–50%
Forecast error reduction with AI-driven forecasting vs. traditional methods
Source: McKinsey & Company
6–15%
Of inventory value released as working capital via optimized safety stock
Conservative estimate · IHL / McKinsey benchmarks
4–8 weeks
To a live pilot on your own SKUs — every feature above included from day one
Defined SKU/location scope

Frequently Asked Questions

Feature questions, answered

What features should demand planning artificial intelligence software have?+
Four things: an ensemble forecasting engine (not a single model), planner workflow with consensus and overrides, inventory optimization that converts forecasts into working-capital decisions, and a governance layer — explainability, bias detection, and audit logs. Demand OS ships all sixteen capabilities in every tier.
Are all sixteen features included, or sold as separate modules?+
All features ship with every tier of Demand OS, including the governance layer. There is no per-module pricing maze; published price bands cover the full capability set.
How does the AI decide which forecasting model to use?+
For every SKU and horizon, the engine evaluates statistical models (ARIMA, Holt-Winters, Croston) and ML approaches (gradient boosting, neural networks) against held-out data, selects the best performer, and logs which model won and why. It re-selects automatically as demand patterns shift.
Can planners override what the software recommends?+
Yes — every forecast, safety-stock target, and replenishment recommendation is editable. Overrides are logged with a reason and timestamp, and Forecast Value Add tracks whether human adjustments are improving the number.

Get Started

See every capability live on your data

One demo covers all sixteen features — forecasting engine, explainability layer, planner workbench, and your integration path — running on your own SKUs. You leave with a scoped pilot plan and published pricing.

Book a Demo