The Data Guyz logo The Data Guyz YOUR DATA EXPERTS Talk to us

SOVEREIGN DATA MIGRATION PLATFORM · ON-PREM, PRIVATE OR PUBLIC CLOUD

Your data is the lifeline. Don’t hand over the keys to move it.

Noviq is an AI data migration platform that runs where your data already lives, on-premise, in a sovereign region, or inside your own tenancy. AI does the discovery and the drafting. Your data stewards approve every decision. The mappings, the code and the IP never leave your estate.

Prove the data layer on one domain before you commit to the big migration, legacy to cloud, big data to lakehouse, or app to app.

Book a 45-minute Noviq demo How sovereignty works

80%

faster onboarding of a new source

0

artefacts generated without a steward’s approval

0

bytes of your data leaving your boundary

100%

of objects converted at Coal LSL, in three months

Migration paths Noviq Shift has conversion coverage for. Hover to pause.

SOVEREIGNTY BY DEFAULT

The engine comes to your data. Your data does not go to the engine.

Most AI migration tooling asks you to ship your estate somewhere else to be understood. For a government agency, a super fund or an insurer that is the end of the conversation. Noviq inverts it: the engine deploys inside your boundary, and only structural metadata ever crosses the line.

WHAT CROSSES THE LINE: AND WHAT NEVER DOES YOUR SOVEREIGNTY BOUNDARYon-prem, sovereign region, or your own tenancy Your source systemsOracle, SQL Server, Hadoop, files, appsNoviq engineProfiling, inference, drafting. Runs in your estateSteward approvalA named person signs each decision ARTEFACTS LAND IN YOUR REPOSITORIES Model endpoint Your own key, your chosen region Swappable. Nothing is hard-wired to one vendor. schema & metadata drafted mappings, returned for review Customer records never cross Column names, types and profiling statistics leave the boundary. Rows do not. If your obligation says nothing may leave at all, Noviq runs against a model inside the boundary instead.
The dashed box is your control boundary. Everything that produces or approves an artefact sits inside it.

So what: you can start a migration programme before your sovereignty review finishes, because there is nothing in the architecture for it to object to. That is usually six to twelve weeks back on the plan.

Deployment topology is confirmed per engagement. Where an obligation prohibits any egress, the model runs inside the boundary and the metadata path above is removed entirely.

THE MIGRATION PATHS

Prove the data layer on one domain before you commit the programme.

The expensive mistake is committing to a full migration before anyone has established what converts mechanically, what needs a human, and what will not reconcile. Noviq is how you find that out on one domain, in weeks, with the evidence to price the rest.

ONE PLATFORM, FOUR MIGRATION SHAPES Legacy → cloudFROMTeradata, Oracle, SQL ServerwarehousesNOVIQ SHIFT DOESObject inventory, SQL andstored-procedure conversion,reconciliationDELIVERED · Coal LSLBig data → lakehouseFROMHadoop, Hive, legacy Spark estatesNOVIQ SHIFT DOESJob refactoring, equivalencetesting, medallion pipelinesTRANSFERABLE PATTERNApp → appFROMOperational systems and their datacontractsNOVIQ SHIFT DOESSchema mapping, field-level lineage,cutover validationTRANSFERABLE PATTERNOn-prem → sovereignFROMRegulated estates with residencyconstraintsNOVIQ SHIFT DOESIn-boundary deployment, stewardapproval, evidence packTRANSFERABLE PATTERN Same engine, same approval gate, same evidence pack. What changes is the conversion coverage and the reconciliation design. Delivered means we have shipped it for a named client. Transferable pattern means the method is proven and the conversion coverage is scoped per estate.
Each column is a migration shape. The coloured rule is the conversion-risk weighting, darker means more business logic embedded in code.

So what: buy the first domain as a fixed-price proof. If the reconciliation holds, you now have a priced, evidenced plan for the whole estate. If it does not, you found out for the cost of one domain.

THE STEWARD MODEL

AI services, supported by your data stewards.

The reason AI migration tooling stalls in regulated estates is not accuracy. It is that nobody can say who approved a decision the model made. Noviq assigns every class of decision to a named human role and records the outcome.

WHO DECIDES WHAT: AND WHAT IS RECORDED Data stewardOwns the business meaningApproves names, types, keys,sensitivityData engineerOwns the implementationApproves mappings,transformations, pipelinesPrivacy / riskOwns the obligationApproves classification, masking,retentionNoviqOwns the draftingProposes, explains, records never decides Every approval writes a decision record: who approved it, what the recommendation was, theconfidence attached, and what they overrode. That record is the audit trail your reviewer asks for. AI services, supported by data stewards, not the other way round.
Noviq proposes and explains; three human roles decide. The dark bar is what gets written every time one of them clicks approve.

So what: when your auditor asks who decided that this column was not personal information, the answer is a person’s name and a timestamp, not “the tool classified it”.

TWO PRODUCTS, ONE OPERATING MODEL

AI recommends. People validate. Nothing ships unapproved.

Noviq

AI-based data onboarding

For teams onboarding new sources. A retail analytics team must onboard a dozen new supplier and e-commerce feeds each quarter into Fabric to support demand forecasting. Each feed takes two to three weeks of manual profiling, mapping and governance review, so the forecasting model keeps slipping.

BEFORE

Engineers hand-profile each file, guess data types and keys, write mapping docs, then rework after governance finds PII.

WITH NOVIQ

AI discovers schemas, proposes types, partitions and candidate keys, and drafts the mapping and technical docs; the analyst validates in a review session, then pipeline artefacts are generated.

OUTCOME

Feeds land in days, governance is applied up front, and the backlog clears, onboarding effort down by up to 80%.

Noviq Shift

AI-based data migration

For teams moving an estate. A platform team must move a SQL Server estate, hundreds of tables, views, stored procedures and functions, to a cloud database before a licence renewal, with no in-house conversion skills and an auditor asking for proof that logic was not altered.

BEFORE

Objects converted by hand, dependencies discovered late, and reconciliation is a spreadsheet exercise that slips the cutover.

WITH SHIFT

The estate is analysed for objects and dependencies, SQL / ETL logic is converted with compatibility gaps flagged for engineer review, and reconciliation reports are produced at every step.

OUTCOME

Coal LSL pattern: 100% of objects converted, validated and deployed in three months with audit-ready reconciliation.

THE OPERATING MODEL

Raw data
AI discovery
Human validation
Artefacts generated
Governed outcome

Nothing is generated until a named person approves it. Artefacts are committed to your repositories, in your stack, under your platform licences.

HOW NOVIQ RUNS

AI does the drafting. A person does the deciding.

Business meaning cannot be inferred safely from column names. Noviq automates the two steps where that does not matter, inspection and generation, and puts a human gate between them, where it does.

THE APPROVAL GATE: NOTHING IS GENERATED UNREVIEWED AI01Inspect source filesFormat, structure, schema andprofiling signalsHUMAN02Review the blueprintTable descriptions, types, keys,partitions, sensitivity labelsHUMAN03Approve the choicesRecommendations, confidence levelsand overridesAI04Generate the assetsMappings, documentation, SQL andpipeline artefacts AI recommends. A named person validates. The decision is captured, so the next auditor can see who approved what. Up to 80% less onboarding effort, because the drafting is automated, not the deciding.
Steps 02 and 03 are the gate. Remove them and you have a code generator nobody will sign off; keep them and the speed still lands, because the slow part was never the typing.

So what: the saving comes from the first and last steps, and the trust comes from the two in the middle. A tool that automates all four is faster on a demo and unusable in a regulated estate.

WHY TEAMS BRING IT IN

Business needs, automation, innovation and platform neutrality.

Business needs

Trusted data for every BU, faster

WHAT TEAMS NEED

  • Onboard new sources in days, not weeks
  • Prove reconciliation to auditors and regulators
  • Free senior engineers for higher-value work

HOW NOVIQ DELIVERS

  • Onboarding blueprint, mapping and documentation generated
  • Audit-ready evidence pack produced per load
80% faster onboarding

Automation

Remove repetitive engineering, keep accountability

WHAT TEAMS NEED

  • Schema discovery, profiling and mapping are manual and inconsistent
  • SQL and ETL conversion is hand-written and error-prone
  • Teams need AI with a human sign-off

HOW NOVIQ DELIVERS

  • AI discovery with mandatory human validation and override
  • Artefacts generated only after approval
0 code generated without review

Innovation

Reuse delivery IP instead of rebuilding

WHAT TEAMS NEED

  • Every project re-creates the same frameworks
  • Knowledge sits in people, not products
  • AI capability must be practical, not experimental

HOW NOVIQ DELIVERS

  • Client engagements productised into Noviq and Noviq Shift
  • LLM-powered metadata generation (Anthropic Claude partner)
4 engagements behind the product

Platform

Neutral, owned, no lock-in

WHAT TEAMS NEED

  • Estates span Fabric, Snowflake, Databricks, AWS and GCP
  • Vendor tools lock the destination to one platform
  • Licence sprawl and catalogue overhead

HOW NOVIQ DELIVERS

  • Runs inside the client's existing stack; outputs committed to client repositories
  • No new platform licence or subscription
5 platforms supported

Partner so-what: a partner can run Noviq inside any client stack, keep the platform relationship it already has, and hand its run team the evidence pack, automation and innovation without a new vendor agenda.

WHAT YOU GET

Artefacts, not a black box.

Noviq is delivery IP we run inside your environment, not a SaaS product you rent. Everything it produces is yours on day one, readable, reviewable and committed to your repositories.

  • 1Source profile and proposed schema, with types, partitions and candidate keys
  • 2Source-to-target mapping and technical documentation
  • 3Object and dependency inventory, with compatibility gaps flagged for review
  • 4Generated pipeline and conversion code, committed to your repositories
  • 5Reconciliation reports and an audit-ready evidence pack per load or wave
  • 6A record of every AI proposal and who approved it

THE FRAMEWORKS ALONGSIDE

What keeps the data trustworthy once Noviq has landed it.

Noviq gets data in and across. These four frameworks, the rest of our delivery IP, keep it defensible afterwards. Each came out of a delivered engagement and is handed over as code you own.

DQ & Load Audit Framework

A configurable, SQL-native data-quality framework with real-time monitoring, plus a load-audit layer comparing source and target record counts across bronze, silver and gold.

  • Rules in SQL your team reads, owns and extends
  • No additional licence, no proprietary rule language
  • Hundreds of rules inside a single run window

MetLife: 99.9% integrity and auditability

ADDF: disposal & de-identification

A scheduled compliance trigger routes end-of-life data to automated purge or de-identification, with Power BI tracking of every disposed and masked record.

  • Compliance becomes a scheduled job, not an annual scramble
  • Primary and cascading tables handled together
  • Analytics value kept without keeping the PII

The Smith Family, compliant by default

AI-Ready Knowledge Foundation

Parallel ingestion pipelines: Zendesk via ADLS and an external stage, SharePoint via Snowflake Openflow, feeding one governed Cortex Search index.

  • Native platform ingestion, no custom integration layer
  • Permissions respected at retrieval, not bolted on after
  • LLM-generated metadata on onboarded tables

The Smith Family, one search across two systems

Customer Data Activation Layer

The warehouse-side half of a CDP or engagement programme: identity resolution, consent and suppression, and audience pipelines into platforms such as Braze.

  • Audience definitions live in the warehouse, not a marketing tool
  • Every send reconcilable back to the record that triggered it
  • Consent applied before activation, not after

Available as a fixed-scope blueprint

WHERE IT GETS USED

Eight tasks teams delegate to Noviq.

Full use-case library
Noviq Shift

Legacy warehouse to cloud conversion

Oracle, SQL Server or Teradata moved object by object, with every table, view, stored procedure and function validated against source.

Coal LSL: 100% of objects in 3 months

Noviq Shift

Hadoop / Hive job modernisation

Getting off an end-of-support cluster without rewriting every job by hand; Hive and MapReduce logic converted with gaps flagged.

Conversion assurance pattern

Noviq Shift

Legacy ETL re-engineering

Hundreds of hand-built Informatica, SSIS or DataStage mappings replaced with a metadata-driven framework your team can maintain.

MetLife, metadata-driven Medallion

Noviq

Quarterly supplier and partner feeds

A dozen new feeds a quarter profiled, mapped, documented and governed in days rather than weeks each.

Up to 80% faster onboarding

Noviq

New source onboarding at scale

CSV, delimited, JSON, XML, Avro and Parquet discovered and standardised into one onboarding pattern across projects.

Woolworths, configuration-driven onboarding

Noviq

Column metadata and DQ checks

Column-level descriptions and candidate quality checks drafted for newly onboarded tables, approved by an engineer before they go live.

Woolworths: LLM metadata agents

Noviq Shift

Core system migration

Customers moved between billing, policy or lending platforms in waves, with staged validation before production approval.

Optus, RESIMAC patterns

Noviq Shift

Stalled migration rescue

An in-flight programme that has slipped twice: object inventory rebuilt, dependencies mapped, reconciliation put back on a footing.

Coal LSL, recovered and landed on time

Platform-neutral, by design.

Noviq runs against whatever you already license. It has no preferred destination, because we don't earn a margin on the one you pick.

TYPICAL SOURCES

Oracle · SQL Server · Teradata · Hadoop / Hive · Informatica · SSIS · DataStage · flat files · REST APIs

TYPICAL TARGETS

Snowflake · Databricks · Microsoft Fabric · BigQuery · AWS (PostgreSQL, Redshift) · Dremio

FILE FORMATS

CSV · delimited · JSON · XML · Avro · Parquet

ORCHESTRATION

Azure Data Factory · Matillion · dbt · Databricks Workflows · Airflow / Composer · Openflow

HOW IT COMPARES

How Noviq compares to the alternatives.

Consideration Noviq / Noviq Shift Vendor-native converters Catalogue / ETL suites Manual / contractor
Platform-neutral target Yes, five platforms Locks to one destination Partial Depends on the individual
Human-in-the-loop by design Mandatory sign-off Optional review Varies Yes, but slow
Client-owned artefacts, no new licence Yes Tied to the vendor Subscription required Yes
Reconciliation evidence built in Yes, at every step Limited Add-on Spreadsheet
Runs inside your existing stack Yes Vendor environment Separate platform Yes

Characterisations describe each category's design intent, not a benchmark of any named vendor.

Common questions

Is Noviq a product we have to buy?

No. It's delivery IP we bring to an engagement and run inside your environment. There's no licence, no subscription and nothing to renew, you pay for the engagement, and you keep what it produces.

Does our data leave our tenancy?

The engagement is designed to run inside your Australian environment, and we scope exactly what the model sees before anything starts. We'll walk your security team through it in the demo.

What if the AI gets a mapping wrong?

It will, sometimes, which is why nothing is generated before a named person approves it. Compatibility gaps and low-confidence proposals are flagged for engineer review rather than quietly guessed.

How do we try it without a big commitment?

A 45-minute demo runs it against a sample of your own schema. A four-to-six week proof of value then puts it against one real source or estate, with an evaluation record you can take to a board.

See it run on your own schema.

Not a canned dataset, a sample of your own. Forty-five minutes, no fee, and your security team is welcome on the call.

Book a Noviq demo See the frameworks