The autonomous operational layer for manufacturing

The system that's supposed to tell your planners what to do —
and doesn't.

Your ERP records what already happened. Your planning tool shows what's happening now. Between them is a human deciding what should happen next, six hundred times a day. Krateos is the autonomous operational layer that runs in that gap, inside your ERP, and makes the decisions the software has never been able to make on its own.

Layer 1 · System of record
Your ERP
Records every transaction. Tells you what already happened.
SAP · Oracle · IFS · NetSuite · Microsoft Dynamics
Layer 2 · Operational decisions
Krateos
Decides what should happen next. Drafts the action. Waits for the nod.
8 agents · live in production · audit-trailed
Layer 3 · Visibility
Your planning tool
A dashboard. Tells you what's happening now.
RELEX · Blue Yonder · Kinaxis · o9 · ToolsGroup
← reads
writes →
The problem nobody solved

Every manufacturer in the world has the same supply-chain problem.

Your ERP records every transaction in your business. Every purchase, every shipment, every invoice. It tells you what already happened.

Your planning tool, sitting next to it, is a dashboard. It tells you what's happening now.

Between them is a human. Your senior planner. Looking at both, looking at the market, looking at the customers, deciding what should happen next. Six hundred times a day. Every day.

When that planner is good, your business is good. When that planner is tired, your business is tired. When that planner retires, the company pays for it for years.

None of this is a problem your team can fix by working harder. They are already working as hard as human beings can work.

It is a problem with the system that sits between your ERP and your planners — the system that's supposed to tell them what to do, and doesn't.

That gap is where Krateos lives.
The mental model

Two analogies. One product.

We built a team of autonomous agents that run inside your ERP and do the work the software has never been able to run. Today there are eight. Two of them carry the load.

01 · Inventory
Walt
Warehouse & Inventory
Walt is the thermostat in your house.

He watches your inventory. When the level drops, he reorders. When it's fine, he doesn't.

His math doesn't need a forecast to work — he runs on reality. Net Flow Position, FEFO-aware on-hand accounting, red/yellow/green zones computed live from MOQ, variability, and lead-time.

02 · Demand
Dana
Demand Planner
Dana is the weather app on your phone.

She looks at patterns and predicts what's coming. Cold snap tomorrow. Below freezing by 6 a.m.

She doesn't replace Walt — she informs him. 19 forecast models per SKU per cycle. Picks the MASE winner. Ships the reasoning stapled.

You don't replace your thermostat with the weather app. But knowing the weather, you might pre-heat the house tonight.

That's what Dana does for Walt. "I see a 50% demand increase coming on RESIN-4200 in June. High confidence. Pre-heat." Walt temporarily raises the inventory target — say, from 24 barrels to 36 — starting in mid-May. When the June surge arrives, the larger buffer absorbs it. In September, when the season ends, the target returns to baseline.

Every vendor in your market has both pieces. None of them run them in conversation.

They export the forecast to a CSV. They import the CSV into the DDMRP configuration. They run the cycle once a month. They hope nothing has changed since.

The CSV dance
How everyone else does it
  • 01Forecasting tool generates a forecast. Once a month.
  • 02Planner exports a CSV.
  • 03Planner imports the CSV into the DDMRP config.
  • 04DDMRP recomputes buffers. Hopes nothing has changed.
  • 05Reality moves. Forecast is stale within days.
  • 06Buffer reacts to actual demand. Forecast doesn't know.
  • 07Repeat next month. Which number do I trust?
The continuous conversation
How Dana & Walt do it
  • 01Dana refits forecasts nightly. Per SKU. Best-of-19.
  • 02Forecast change pushed to Walt. Within seconds.
  • 03Walt applies the Demand Adjustment Factor. Pre-heats or cools.
  • 04Buffer reacts to actual demand in real time.
  • 05Walt's reaction observed by Dana. Models recalibrate.
  • 06Both write to the same audit trail. One source of truth.
  • 07No CSV. No monthly cycle. One number to trust.

That continuous conversation is the product.

A night in production

Eight agents. One audit trail. A single shift.

Dana and Walt carry the load. Six others — Brian, Quinn, Sarah, Max, Frank, Grace — pick up the work that follows. By 7 AM there are six decisions in your inbox. Not six hundred.

02:58
Dana Demand Planner
Forecasts while you sleep. 19 models benchmarked per SKU. Picks the MASE winner. Refits on full history. Ships with archetype, CV window, and SHAP importances when ML earns the pick.
audit: 2,847 forecasts · 12m 14s · hierarchy reconciled
03:15
Walt Warehouse & Inventory
Catches the buffer breach. Net Flow Position on RXN-4401 dropped below top-of-red, aged 3 days. Flags severity, surfaces lead-time, pings Brian with context — and Dana's reasoning attached.
audit: zone=RED · NFP=847 · TOY=1,340 · TOR=1,010
03:30
Brian Procurement
Drafts the PO. MOQ-aware, supplier-scored, priced against last quarter. Confidence-gated: auto-propose if MASE < 1.0, else manual review.
audit: PO-88214 · ACME-Poly · 1,200u · $18,420 · awaiting approval
03:48
Quinn Quality & Compliance
Catches the expiring lot. FEFO-aware on-hand accounting finds 80 units of L-22991 aging out in 14 days. Reroutes allocation, caps buffer sizing on that SKU to shelf-life ceiling.
audit: lot L-22991 · expires 2026-05-08 · reallocated to order 77-041
04:01
Sarah Customer Service
Drafts the customer email before the customer wakes up to write it. Delivery-window shift, new ETA, apology calibrated to account tier, CC the right rep.
audit: customer 11-221 · draft saved · not sent · awaiting approval
04:14
Frank Accounts Payable
Closes the three-way match. PO + receipt + invoice tie out within tolerance. Books the liability. Files the exceptions that don't.
audit: INV-552312 · matched · variance $0.00 · posted to GL
04:30
Max Production Ops
Reshuffles the production schedule. Swaps job 8812 with 8814 to honor buffer priority and keep the changeover penalty below its tolerance.
audit: schedule v142 · 3 jobs moved · changeover Δ 18min
05:02
Grace Accounts Receivable
Chases the AR. Account 44-118, 36 days past due. Second reminder, tone a notch firmer. Escalation path queued for the human if it doesn't resolve by Friday.
audit: 4 reminders sent · 2 disputes escalated · $187,240 outstanding
The roster

Two leads. Six specialists.

Dana and Walt run the operational core — forecast and replenishment, in continuous conversation. The rest pick up the work that follows. Each named, role-scoped, and authority-limited. Trust widens only as they earn it.

Dana Lead
Demand Planner
Forecasts 2,800+ SKUs nightly across 19 models — picks the MASE winner, ships reasoning stapled.
01 / 08
Walt Lead
Warehouse & Inv.
Watches Net Flow Position against red/yellow/green. Catches breaches before they become stockouts.
02 / 08
Brian
Procurement
Drafts POs with Dana's reasoning attached. Supplier-scored, MOQ-aware, priced against history.
03 / 08
Quinn
Quality & Compliance
FEFO-aware lot tracking. Catches expiring inventory. Caps buffers at shelf-life ceilings.
04 / 08
Sarah
Customer Service
Drafts customer comms before the customer drafts theirs. Tone-calibrated to account tier.
05 / 08
Max
Production Ops
Reshuffles the production schedule against buffer priority and changeover tolerance.
06 / 08
Frank
Accounts Payable
Three-way match, clean. Books the tie-outs. Files the exceptions that don't.
07 / 08
Grace
Accounts Receivable
Chases aging AR. Firm but on-brand. Escalation path queued for the human if needed.
08 / 08
Proof · Dana

The forecasting depth no one else ships.

19 models benchmarked per SKU per planning cycle. SeasonalNaive is the floor — no forecast ships unless it beats naive on hold-out MASE.

Classical stats

Eight models. One winner.

  • autoAutoARIMA · AutoETS · AutoTheta · DynamicOptimizedTheta
  • seasonalMSTL · Holt-Winters · SeasonalExponentialSmoothing · SeasonalNaive
  • selectionHold-out MASE bake-off. Refit on full history.
Intermittent

Low-velocity SKUs don't get faked.

  • familyCrostonSBA · CrostonClassic · ADIDA · IMAPA · TSB
  • routed5 archetype classes pick the family. Trending · seasonal · multi-seasonal · intermittent · spiky.
  • gateMust beat naive on MASE. Else the forecast doesn't ship.
Machine learning

LightGBM, three paths.

  • per-SKULag features [1, 4, 13, 52] + exogenous regressors.
  • batchShort-history SKUs borrow strength from similar ones — the primary ML advantage.
  • bake-offHead-to-head stats vs. ML. MASE winner picked. No shrugging.
Reconciliation & explainability

The sum of the parts equals the whole.

  • hierarchyCustomer × Region × SKU reconciled via Nixtla's open-source solver.
  • shapSHAP feature importances when ML wins. You know why.
  • shippedArchetype · MASE · CV window · importances. No black box.
What ships with every forecast{ archetype: "seasonal", mase: 0.74, cv_window: "8w rolling", winner: "LightGBM-batch", shap_top3: ["lag_52", "promo_flag", "lag_4"] }
Proof · Walt

DDMRP, canonical — not "inspired by."

Demand-Driven Material Requirements Planning, implemented to the spec. Red / yellow / green from MOQ, variability, lead-time, and cycle factor. Live today across 31 SKUs.

Green
Yellow
Red · 3d aged

Net Flow Position, FEFO-aware. What's actually available after allocation and expiration — not what the ERP thinks is in the rack.

  • Three profile archetypes · LONG-HIGH-MOQ, MED-MED-MOQ, SHORT-LOW-MOQ
  • Demand Adjustment Factor — symmetric. Saves by not buying.
  • γ-reprofile — canonical profile-clone-and-reassign. No parallel override files.
  • Shelf-life ceiling cap — no buffer bigger than what will consume before expiration.
  • Aging-in-red — not "in red" but "in red for N days." Severity operators can prioritize.
  • Confidence-gated auto-propose — bad signal never becomes an auto-action.
  • Atomic writes. Overlap rejection. Baseline frozen at approval. Audit integrity by construction.
Portability

Your agents. Your playbooks. Your data. Your cloud.

Agents
Your org chart.
Named, scoped, configured to your authority limits. Not ours.
Playbooks
Your playbook.
Written in English. Versioned in code. Diffable in git.
Data
Your database.
Lives where your ERP lives. We read; you keep.
Cloud
Your tenant.
VPC, BYOC, on-prem. The agents run where you say they run.
What's next

The gap is where Krateos lives.

hello@krateos.ai · Tonight Dana refits 2,847 forecasts. Walt will catch the breach around 3 AM. Brian will have the PO drafted by 3:30. By 7 AM there will be six decisions in your inbox.