InRecipe

Industrial item data is full of blanks.

Missing weights and dimensions, no customs code, manufacturer names written five different ways. The blanks stop procurement, pricing, item creation and customs handling, and filling them by hand is months of one person's time.

Filling them automatically is easy. Filling them correctly is harder: a wrong number looks exactly like a right one, and the only thing that tells them apart is the source it came from.

What InRecipe is

InRecipe completes item, parts and equipment records against a curated source library built specifically for industrial equipment and parts, indexed per industry and kept current, so a value can be traced to where it came from rather than inferred.

Generation and verification are separate on purpose. A validator checks every proposed value against its citation before it is written into the record, so nothing grades its own work. Every field comes back one of three ways.

A verified value

Taken from a manufacturer source, with a citation that points back to it.

An honest gap

Searched, and nothing trustworthy found. A real blank, not a guess.

!

A flagged estimate

Bounded, never disguised as fact, and routed to a person to review.

Enriched data is delivered in a form your operational systems take in: ERP, PDM, PIM, CMMS. Getting started does not require an integration project.

What complete data unlocks

Complete item data is rarely the goal in itself. It is what the goal depends on.

01

Procurement and sourcing

Supplier rows become items the ERP accepts, with the evidence attached before anything is created. Once components are described consistently, near-identical items become visible across systems. Spend can be consolidated, and specifications are complete enough to search for an alternative when a component reaches end of life.

02

Pricing, planning and analytics

Models and reports stop inheriting blanks and five spellings of the same manufacturer. Forecasting and contract pricing improve measurably on the same model, with no change other than the input.

03

Customs and governance

Every value is dated, sourced and attributable, and an estimate is marked as an estimate. An audit reads the record instead of asking people what they remember.

Four case studies

Item masters, spare parts, a device register and supplier price lists. Each case was run on a real industrial company's own data, taken from its own systems. The companies are kept anonymous, and every figure below comes from that work.

Case 1Manufacturing

1,955 items completed, and the existing master corrected

An equipment manufacturer with a product master of 6,000 items across 10 groups, covering mechanical hardware, electrical, piping, HVAC and propulsion.

  • 1,955 items enriched out of the 6,000-item master
  • Every value hand-checked against its source turned out correct
  • Where our value disagreed with the existing record, the existing record was the wrong one in three cases out of four
Case 2Industrial services

Months of hand work done in under three hours

A service company maintaining a register of roughly 2,500 devices on behalf of its business customers.

  • Roughly 2,500 devices, 17 fields filled per device
  • The whole run finished in under three hours
  • Manufacturer names made consistent on more than 95% of records
  • A forecasting model rerun on the enriched data came out 50% more accurate, same model, cleaner input
Case 3Heavy equipment

One parts catalogue enriched, four systems served

An equipment manufacturer with a very large spare-parts catalogue.

  • Parts need a different treatment than devices, so the company's own classification is the reference
  • One enrichment feeds four systems downstream
  • New and changed parts stay in scope, not a one-off cleanup
Case 4Technical distribution

2,736 supplier rows became items the ERP accepts

A technical distributor selling equipment on behalf of its suppliers.

  • 2,736 supplier rows composed into items the ERP accepts
  • An item number generated for every row
  • Changed prices caught and highlighted automatically
  • Checked against a parallel AI run done independently on the same rows
  • One supplier first, with the same pattern repeating for around ten more

Contact us

Leave your email and we'll get in touch to agree a first scan on your own records, and see what's possible.

Or write to us directly: inrecipe@datadesign.fi