Contact us

Smelter · Data harmonization

One number.
Everyone agrees.

One record. Every source. Zero movement.

Book a scoping session How it works

Every team used to hold its own figure. Smelter builds one record from every system, and every team sees the same one.

Before Smelter

Click the ones you recognise. Each one shows what changes.

Tonight, somewhere in your company.
What this adds up to 0 of 6

Smelter is the part that holds the record.

Every source resolved into one governed record, with lineage attached and nothing copied out of your walls.

See how it works

After Smelter

The same six, once Smelter is running.

Click any card to flag or unflag that problem.

Want one number your teams agree on?

Book a scoping session See how it works

How Smelter works

Raw data in. One trusted record out.

1 · Bring it in

Every source, read where it sits. Nothing is moved.

2 · Clean and check

Duplicates are merged. Bad rows wait for a person.

3 · One record

Only checked data goes out. Every team reads the same.

How it is built

MeltanodbtPolarsDagsterApache IcebergTrinoPostgreSQLOry

Open source and open formats. Runs air-gapped, inside your walls.

Open formatsNo lock-inAir-gapped
Where it runsAn air-gapped server room inside a client sitePhoto slots-runs.jpg · 16:9
The record, air-gapped on your own floor.

Industries that use Smelter

And, just as clearly, what it is not.

One number everyone can defend, without moving your data.

Your systems stay where they are. The record is built on top, inside your walls.

Pick your industry

See every industry

Smelter · record Photo slots-what.jpg · 16:9
One record, inside your walls.

What it is

  • A governed record harmonizing internal and external sources, air-gapped in your perimeter.
  • Reporting without copying data, on open formats (Apache Iceberg), queried in place.
  • Bad data quarantined for human approval before it is published.

What it is not

  • Not a cloud data warehouse. It runs in your perimeter.
  • Not a BI or analytics tool. Keep yours on top.
  • Not a dumping ground. Nothing is published until it passes validation.

Case studies

Where Smelter is used. Pick an industry, or let it run.

Industry · Trade

We turned five inboxes into one clearing flow for a trade house

The problem

Each party holds one piece of the picture, and the clearing record is assembled by hand from all of them.

How we solved it

Outside data brought in and joined into one record, then cleared by versioned rules with every step logged.

The result

One path through the process instead of five inboxes.

Read the full story

What was going on

Clearing a single trade meant chasing data from every party involved. Each one held a piece of the picture, and someone stitched the clearing record together by hand from five different inboxes.

What we did

  1. Quarry brought in the outside data each party publishes, in one format.
  2. Smelter joined every piece into one record of the trade, with each value keeping a note of where it came from.
  3. Assembler runs the clearing steps as versioned rules, and writes every step to a log the client owns.

What changed

One path through the process instead of five inboxes. When someone asks why a trade cleared, the answer is already in the log.

All three

Industry · Retail

We gave a retailer one set of numbers

The problem

Sales, stock and supplier systems each hold a version of the truth, and none of them agree.

How we solved it

Every source is read where it sits and resolved into one record, with each value keeping its source.

The result

One set of numbers for the week, with nothing copied out.

Read the full story

What was going on

Sales, stock and supplier systems each reported a different number for the same week. Analysts exported all three into spreadsheets and argued over which one was right, and the meeting started from the argument, not the decision.

What we did

  1. Smelter reads each system where it sits. Nothing is copied into a new warehouse.
  2. Records that describe the same product, store or supplier are matched and resolved into one record.
  3. Every value keeps its source, so anyone can see which system a number came from.

What changed

One set of numbers for the week, with nothing copied out. The team starts from the same figure, and can check where it came from.

Smelter

Industry · Finance

We reconciled two finance firms without sharing their data

The problem

Two organisations have to agree on figures that neither is allowed to share.

How we solved it

Both sides compute over the record without exposing raw data to each other.

The result

Reconciliation without a data-sharing agreement.

Read the full story

What was going on

Two firms had to agree on shared figures at the end of every period. Neither was allowed to show the other its raw data, so reconciliation ran on emailed summaries and a great deal of trust.

What we did

  1. Each firm’s data stays inside its own environment.
  2. Smelter lets both sides compute over the record together, so the figures are compared without either side exposing its raw data.
  3. Where the numbers differ, both sides see the difference, not each other’s books.

What changed

Reconciliation without a data-sharing agreement. The figures are agreed, and the private data stays private.

Smelter

See all case studies

~66% data-team productivity gain reported for model-driven data platforms. Category benchmark, Forrester, 2026. Directional: your own measured results replace it per deployment.

FAQ

The basics of Smelter.

What is Smelter?

Smelter is a data harmonization platform. It brings every source, internal and external, into one governed record: ingested, transformed, validated, approved and published, on an open lakehouse inside your own environment. Every team reads the same trusted numbers, and nothing reaches the record without passing your rules and, where it matters, a person.

What problem does Smelter solve?

Every team holding its own figure. When ERP, CRM, spreadsheets and external feeds disagree, people reconcile by hand and argue about whose number is right. Smelter builds one record from every system, holds bad data for review instead of publishing it silently, and keeps an audit trail of every rule, exception and approval.

Who uses Smelter?

Data engineers build pipelines in SQL and Python; analysts query and build dashboards; data stewards review quarantined rows and approve what is published; platform operators run the environment and onboard tenants. Business teams consume the trusted record through their BI tools and through Assembler.

Is Smelter a data warehouse or a BI tool?

Neither. Smelter is the governed record, not the dashboard: keep your BI tools on top through SQL. It is also not a cloud warehouse: it runs inside your perimeter, on open table formats any engine can read.

What does "data harmonization" mean in practice?

Bringing sources with different shapes, keys and quality into one consistent model: cleaning and typing values, normalising formats such as emails and phone numbers, de-duplicating and merging on business keys, conforming entities across systems and validating the result against your business rules before anyone relies on it.

See all 31 Smelter questions →

The other two platforms

Each is useful alone. Together they carry one piece of data end to end.

Quarry

All your outside data. One clean feed.

See QuarryAssembler

Change a rule. It’s live.

See AssemblerStart here

Forty-five minutes with the people who built it.

Book a scoping session