Request a demoDemo

Why PlanSieve

Don’t buy a tool welded to one problem. Adopt an engine that speaks your business.

This page is the point of view. It names the pattern the category is built on, the pattern we refused, and the principles you can hold us to.

The thesis

The engine claim, and the product claim on top of it

The legacy planning suites fuse their engine to their schema. Item, location and customer are baked into the core, so every new question is a migration and every customer is a multi-year project. PlanSieve refuses that fusion: the data model, the math, the interface, the connectors and even the storage engine are all pluggable behind eight contracts. That is the engine claim.

The product claim sits on top of it. A planner gets a surface where every setting is a switch they own, every number explains itself, and a verdict takes one gesture. The complexity of a business goes in: its dimensions, its history, its drivers, its exceptions. And what comes out is a plan that can be asked why.

Complexity inClarity out

Two patterns

Engine fused to schema, or engine behind contracts

This is a contrast between two ways of building a planning platform, not between two products. The first pattern is the category’s. The second is ours.

FacetEngine fused to schemaEngine behind contracts
The data modelItem, location and customer are baked into the core. A new axis is a migration.Dimensions, levels, hierarchies and grains are data. A new axis is a metadata change.
The mathOne privileged model, chosen at go-live. A new method is a release, or a services engagement.Every model, solver and feature enters through one contract and competes on the same backtest, yours included.
The storageA single engine the product is welded to. Small and large customers get two products.The fact store is an adapter: embedded on a laptop, scale-out in production, with the same model and screens on both.
The interfaceHand-built screens; the API is an afterthought. Automation stops where the screens start.Every screen renders from metadata through the same doors a program uses. Agent-ready by construction.
BehaviourEngine behaviour is a consultant’s configuration job or a support ticket.Every behaviour is a policy: a switch with a dial, set once at the level you choose, off provably skipping.
A new moduleA new product with a new schema and a new implementation.A versioned package of metadata you activate, diff and upgrade. The core does not change.

The platform, chapter by chapter

Eight principles

The principles a buyer can hold us to

These are invariants, not aspirations. A design that violates one is wrong, not creative, and each one is proven on the platform page.

  1. No hardcoded business concept

    There is no item table, no location table, no customer table in the core. If a concept were hardcoded, we would call it a bug. Your vocabulary is the model.

    Proven under Your vocabulary is the model

  2. Everything configurable from the interface

    If changing business behaviour needed a code edit or a config file, it would not be done. Formulas, rules, pages, sets, policies, actions: all screens, all versioned.

    Proven under Everything is a policy

  3. Algorithms are plug-ins

    The built-ins use the same contract your proprietary model would. Nothing is privileged; everything is scored by the same backtest and recorded in the same lineage.

    Proven under Open by design

  4. Storage is an adapter, not a vendor

    One logical model, swappable storage, a columnar interchange in between, so the same product is honest on a laptop and on a billion rows.

    Proven under The engine

  5. Platform first: modules are configuration

    The platform is done when its engines and contracts are complete; a module is a package of metadata. Demand ships on the same core that will carry the rest.

    Proven under Open by design

  6. Nothing is a black box

    Every number is traceable to its inputs, its formula or model, its version, and who or what produced it, and the explanation is of the number you clicked, always.

    Proven under Nothing is a black box

  7. Secure and multi-tenant from row one

    Isolation and permissions are in the first line of the schema. Retrofitting tenancy is a rewrite; we refused to earn that debt.

    Proven under Secure from row one

  8. Truthful by construction

    Accuracy, bias and forecast value added are measured and surfaced automatically, including when a human override made the plan worse.

    Proven under Nothing is a black box

What you will not find here

No logo wall. No uplift percentage. No badge.

A pre-customer product has one credibility asset, and it is not a claim. It is refusing to make the ones it cannot back.

  • No customer proof. No logos, no testimonials, no case studies. None exist yet, and a fabricated one would cost more than it earned.
  • No accuracy uplift. Forecast value added and accuracy are capabilities the platform surfaces, not results we have delivered to anyone. We will not quote a percentage we did not measure on your data.
  • No certification. We describe the security posture, row-level isolation, audit, self-hosted observability with no external egress, and not a certificate we have not obtained.
  • No timeline promise. The platform is designed so a new planning problem is configuration, not a rebuild. We say that, and only that, until we have proven more.
  • Every number carries its framing. Measured in our lab, on our own test data, never a customer benchmark, never a cloud cluster. It says so beside the number. The full table, with caveats.

We write in the company’s voice. What is on this site is what the product does, and every claim names where it was measured.

Judge the engine on your data

Bring your history, your hierarchies and your vocabulary. We show the platform working on them, and where it does not, we say so.

Request a demo

a working demo on your data · no commitment