Use cases

Start from the problem you have

Seven situations that bring teams to VaultSpeed, each with how it runs, what you get and who did it before you.

Book a working session How the platform works

Definitions agreed once

Build an enterprise or canonical model

Much of the enterprise model is already agreed on, in the catalog, in older models and with domain experts, and AI initiatives need it in one place. Writing it by hand takes years.

Two colleagues working through a data model on a whiteboard

How it runs

Start from an industry model

FIBO, ACORD, HL7 FHIR, BIAN, ISA-95, SID or HEDM give the agent a starting structure for your sector.
Start from an industry model

Bring in what is agreed

Definitions from the catalog, existing models and expert interviews are read into the context store.
Bring in what is agreed

Draft and decide with an agent

The agent proposes entities, keys and relationships; your modelers decide, and every decision is versioned in Git.
Draft and decide with an agent

What you get. An enterprise data model held as one graph, that engineers and agents query over MCP.

Convert or draft the data modelBook a working session

Modeling tools out of maintenance

Move off Erwin or PowerDesigner

Erwin, PowerDesigner and ER/Studio hold years of modeling work, and for many teams the tools are out of maintenance. They generate the DDL and stop there: the transformation code is written beside the data model, and the two drift apart.

How it runs

Import the data model

Agents convert the Erwin or PowerDesigner export into the context store, entities, relationships and definitions included.
Import the data model

Keep modeling with AI

Agents propose changes in sessions your modelers review, so the data model stays current.
Keep modeling with AI

Generate the code

The generator writes native code for your platform from the data model, the same way every run.
Generate the code

What you get. A machine readable data model with its lineage and logic, and code that follows it.

How modeling works with an agentBook a working session

Leaving a legacy platform

Rebuild or move the warehouse, generated from the data model

Your warehouse runs because your team made it run: ETL mappings, stored procedures and SQL tuned over years, much of it outside any data model. Moving from Teradata, Oracle or Cloudera to Snowflake, Databricks, Fabric, BigQuery or Redshift means rediscovering, rewriting and retesting all of it.

How it runs

Read the current platform

Agents recover the data model behind legacy code, mappings and schemas into the context store.
Read the current platform

Review and choose the target

Your team checks the recovered data model and lineage, then picks the platform and modeling approach.
Review and choose the target

Regenerate and deploy

Native code for the new platform is generated from the reviewed data model, then reviewed in Git and deployed.
Regenerate and deploy

What you get. Logic that lives in the data model, with lineage from the old platform to the new, so the next move is a regeneration.

Book a working session

Customer story · Retail

Thanks to VaultSpeed, we industrialized our data layer without compromising agility or compliance. Our teams build and deploy data products faster, smarter, and at scale.

Colruyt Group

Colruyt GroupWarehouse rebuild and modernization, from Cloudera to SnowflakeRead the story

Two data landscapes become one

Integrate two data landscapes on shared business concepts

An acquisition brings a second warehouse, a second set of sources and a second definition of customer, product and revenue. Reconciling that in code, one mapping at a time, is slow and depends on the few people who know each side.

A data team working together at one table

How it runs

Read both sides

Agents load data models, code and lineage from each side into one context store.
Read both sides

Decide the differences

Each side is mapped to shared business concepts; where definitions conflict, your teams choose the rule, versioned in Git.
Decide the differences

Generate and consolidate

Integration code is generated from the aligned data model. Keep both platforms running, or retire systems when ready.
Generate and consolidate

What you get. One data model both sides can read, with every integrated figure traced back to its side and source system.

Read how Thomson Reuters did itBook a working session

The source system changes

ERP migration that keeps your business model

A move from SAP ECC to S/4HANA, or to any new ERP, changes the tables and fields beneath your warehouse while customers, orders and products stay the same. It runs in waves, old and new side by side, and every hand written mapping is fixed by hand.

How it runs

Analyze the new ERP

Agents profile the new source structures and propose how each table and field maps to your business concepts.
Analyze the new ERP

Review the impact

See which parts of the data model, loads and data products a change touches before anything is generated.
Review the impact

Regenerate per wave

Integration code is regenerated from the data model, reviewed in Git and deployed, wave after wave.
Regenerate per wave

What you get. A business model that holds steady through the cutover, both ERPs feeding the same concepts, lineage through the change.

Read how Grundfos did itBook a working session

Under regulatory scrutiny

Regulatory reporting you can trace and reproduce

The transformations behind a regulatory report are often spread across hand written scripts, scheduled jobs and the memory of a few engineers. When an auditor asks how a figure was derived, the team reconstructs the answer by hand.

How it runs

Capture the landscape

Sources, data model, lineage and transformation logic sit in one graph, including the code you already run.
Capture the landscape

Change the rule once

The rule is updated in the data model; agents propose the change as a versioned commit your team reviews before it ships.
Change the rule once

Regenerate and run

The code regenerates from the data model and runs in your own cloud, with lineage intact for the next audit.
Regenerate and run

What you get. You can trace and reproduce a reported number: every column has its lineage, every change an author, a review and a history.

See how Argenta did itBook a working session

Customer story · Public sector

VaultSpeed lets us go from data models to impact faster, cleaner, and with more confidence.

Investissement Québec

Investissement QuébecRegulatory and compliance reporting on SnowflakeRead the story

Before the agents scale

Turn an existing Data Vault into a foundation agents can work from

Your teams are already shipping with AI, and nothing holds the context between sessions.

Built a Data Vault by hand or with other tooling? Import it into the context store: your code keeps running, your design stays yours, and agents work from the imported data model.

An analyst working with the data in the evening

How it runs

Describe the landscape once

Agents read your data models, code and requirements into the context store; your team checks what they found.
Describe the landscape once

Point your agents at it

Your agents, coding assistants and VaultSpeed agents read the same store over MCP, so none relearns the landscape.
Point your agents at it

Review what they change

Every proposal arrives as a versioned change in Git, reviewed before it ships.
Review what they change

What you get. AI work with a memory: what one session learns stays in the store, and the next one starts from it.

Read how Humana Healthcare Research did itBook a working session

Talk to us

Start with the situation you are in

Tell us which of these you are facing and what you run today. We come back with a proposal for a first working session on your own metadata.