Informat Limited · Hamilton, New Zealand

Raw data in. Trusted decisions out.

We take organisations from data to information to intelligence. We build the platforms New Zealand organisations run on, then the analytics, AI and machine learning products that sit on top of them.

Describe your data challenge What we do

SourcesERP · CRM · APIs

Bronzeraw, as-landed

Silvercleansed, conformed

Goldmodelled, governed

ProductsBI · AI · APIs

Consumerspeople · apps · AI agents

pipeline: customer_360rows processed: 0dq checks: passing

Our vision

Data that every team can trust and use.

We want every organisation we work with to have one source of truth, delivered fast, governed well, and ready for AI.

Trust

Governed, tested and documented data with clear lineage from source to report.

Speed

Modern cloud platforms and automation that shorten the path from question to answer.

Intelligence

AI and machine learning built on solid foundations, not on spreadsheets.

About us

A specialist data consultancy that builds, not just advises.

Informat Limited is a New Zealand data services company. Our consultants have delivered data platforms and migrations inside banks, government agencies and SaaS companies, and we bring that hands-on experience to every engagement.

DATA

Engineering & platforms

Reliable pipelines and a modern platform to land every source.

Lakehouse & warehouse build
Batch & streaming pipelines
Legacy-to-cloud migration
Master data management
DataOps & CI/CD

INFORMATION

Modelling & insight

Turn raw tables into trusted, well-defined business information.

Kimball & Data Vault models
Semantic layer & metrics
Power BI & self-service
Catalogue, lineage & quality
Data governance

ARTIFICIAL INTELLIGENCE

Generative AI & agents

Put large language models to work on your own governed data.

Chat with your data (RAG)
Natural language to SQL
Document intelligence
AI agents & automation
Private, on-prem LLMs

MACHINE LEARNING

Prediction & MLOps

Models that forecast, score and detect, running in production.

Demand & revenue forecasting
Churn & propensity scoring
Anomaly & fraud detection
Explainable AI
MLOps with MLflow

Secured end to end. Our ICT Security & GRC practice covers access control, privacy (Privacy Act 2020), responsible AI and audit-ready lineage across every layer.

BankingDelivery inside NZ retail banks

Public sectorFabric migration for a Crown entity

SaaSLarge-scale platform migration

NZ-ownedBased in Hamilton, working nationwide

Technology

We work in the platforms you already own.

Platform-neutral advice, deep hands-on skill. We pick the stack that fits your team, budget and data, not the other way round.

Platforms

DatabricksSnowflakeMicrosoft FabricAzure Synapse

Engineering

Azure Data FactorydbtPySparkSQLPython

Modelling & MDM

KimballData VaultMedallionProfisee

AI & analytics

Power BILLMs & RAGMLflowPrivate LLM Lab

AI & ML Lab

Try it before you talk to us.

Five working demos of what we build for clients. Each runs in your browser on sample data.

The same data, four layers up

Every engagement climbs this ladder. Pick a layer to see one week of retail transactions from a Waikato chain at that stage.

Sample banking schema

customers(customer_id, name, region, segment, joined_date)

accounts(account_id, customer_id, product, opened_date, balance)

transactions(txn_id, account_id, txn_date, amount, channel)

Business users ask questions in plain English. The model writes Databricks SQL against your governed tables and explains it.

Our products

Built by Informat.

Alongside client work, we build our own data technology.

Ryu

RyuDB

v0.1 · preview

A GPU-powered SQL database. RyuDB runs analytical queries on NVIDIA GPUs and takes writes through a transactional layer, so one engine serves both analytics and updates (HTAP).

GPU query engineSQL executed on NVIDIA GPUs with RAPIDS cuDF and custom fused C++/CUDA kernels.

HTAP storageColumnar Parquet base with a delta store, write-ahead log and MVCC transactions.

Postgres compatibleConnect with psql, psycopg, asyncpg or JDBC over the Postgres wire protocol.

Rich SQLJoins, window functions, subqueries, ROLLUP/CUBE and INSERT, UPDATE, DELETE and MERGE.

Web consoleWorksheets, query plans, charts, dashboards, notebooks and natural-language-to-SQL.

Ships in DockerOne image with the CUDA runtime. Run it on any host with an NVIDIA GPU.

View on GitHub Talk to us about RyuDB

psql · ryudb on RTX 3090

$ docker run --gpus all -p 5432:5432 ryudb
$ psql -h 127.0.0.1 -U ryudb ryudb

ryudb=> CREATE TABLE lineitem FROM '/data/lineitem/*.parquet';
ryudb=> SELECT n_name, SUM(l_extendedprice) AS revenue
          FROM lineitem
          JOIN orders   ON l_orderkey = o_orderkey
          JOIN customer ON o_custkey  = c_custkey
          JOIN nation   ON c_nationkey = n_nationkey
         GROUP BY n_name;
(25 rows)  ·  60M-row fact table  ·  fused GPU kernel
Time: 64 ms   DuckDB: 412 ms

Speed-up over DuckDB, TPC-H scale factor 10

Warm path (data resident on the GPU), NVIDIA RTX 3090. Results verified against DuckDB.

4-table join + aggregate6.49×

Q6 filter + global agg2.77×

Q1 pricing summary1.97×

Scan + 5 aggregates1.76×

High-cardinality group-by1.74×

Q3 3-table join + agg1.45×

1× DuckDB3×5×7×

Industries

Data and AI for the sectors that run Aotearoa.

Each industry has its own data, its own regulations and its own best first win. Pick yours.

The challenge

Legacy core-banking extracts and overnight batch jobs
Regulatory and industry reports with manual steps
Reconciliations still done in spreadsheets
Customer data spread across many systems

What we build

Lakehouse on Databricks with Unity Catalog
Automated reconciliation and exception reporting
Reports migrated off Tableau and MySQL
Churn, propensity and fraud-detection models

Example data product

Daily payments exception report

Matches every transaction in a scheme settlement file against the ledger on six keys, enriches it with customer details and flags breaks in Power BI each morning.

Designed for:Privacy Act 2020RBNZ outsourcing policy (BS11)CoFI conduct regimePCI DSS

Our work

Selected engagements.

Recent platform, reporting and integration work for New Zealand organisations.

Public sector

Legacy analytics estate moved to Microsoft Fabric

Challenge: A Crown financing entity reported from Synapse, Dataverse and Power BI, while an 87-table Azure SQL system of record was being introduced.
Approach: Full medallion architecture in Fabric. Self-discovering, schema-drift-aware ingestion notebooks with Entra ID auth. Field-level mapping of a 40-table semantic model and around 466 DAX measures to the new source. Private endpoints and automated provisioning with Python and MSAL.
Stack: Microsoft FabricPySparkAzure SQLPower BIEntra ID

One governed Fabric platform, backed by a 17-page architecture design

Banking

Automated payments reconciliation

Challenge: Card-scheme settlement files had to be checked against a suspense account by hand to find breaks.
Approach: PySpark reconciliation in Databricks matching on six keys (RRN, STAN, amount, date, time and message type), with customer details linked through card and account records. Exceptions published to Power BI.
Stack: DatabricksPySparkDelta LakePower BI

A daily exception report designed to replace manual checking

Banking

Industry reporting moved off legacy tools

Challenge: A regulated industry report ran on Tableau and MySQL, outside the bank's strategic data platform.
Approach: Rebuilt the report logic in Databricks against agreed business rules, such as a 13-month window for direct debits. An interim approach kept delivery moving while the replacement source pipelines were built.
Stack: DatabricksSQLPower BITableau migration

Report rebuilt on the strategic platform with the MySQL dependency removed

Banking

Marketing data on autopilot

Challenge: Marketing activity data sat in Marketo and was exported by hand for analysis.
Approach: Automated ingestion from the Marketo Bulk Export API into Databricks, with credentials held in managed secret scopes.
Stack: DatabricksPythonREST APIsSecret scopes

Marketing data lands in the lakehouse automatically, ready for customer analytics

Client names are withheld to respect confidentiality. References are available on request.

Our team

Senior people on every engagement.

Amir

FOUNDER · PRINCIPAL DATA CONSULTANT

Senior data engineer and architect with delivery across Databricks, Snowflake, Microsoft Fabric, Azure Data Factory and dbt. Has led platform work in NZ banking, government and SaaS.

ICT Security Specialist

JOINING SOON

Leading our new Security and GRC practice: data protection, access control and compliance frameworks.

Associate network

ON DEMAND

Trusted specialists in BI, analytics and AI engineering who join larger programmes when you need extra capacity.

AI data advisor

Tell us what's slowing your data down.

Our advisor asks about your situation, suggests how Informat could help, and sketches a proposed data platform for you as an animated diagram. When you're ready, hand the conversation to Amir on WhatsApp.

The advisor gives general guidance only. A consultant will confirm any recommendation before you commit.

Informat Advisoronline

Kia ora! What data challenge are you facing? For example: slow reports, a legacy warehouse you want to retire, or wanting to use AI on your own data.

Want to talk to a person?Send this chat to WhatsApp →

Contact

Start with a conversation.

Message us on WhatsApp for the fastest reply, or leave your details and we'll be in touch within one business day.

CompanyInformat Limited

LocationHamilton, Waikato · working across Aotearoa

Webinformat.co.nz

Raw data in. Trusted decisions out.

Data that every team can trust and use.

Trust

Speed

Intelligence

A specialist data consultancy that builds, not just advises.

Engineering & platforms

Modelling & insight

Generative AI & agents

Prediction & MLOps

We work in the platforms you already own.

Platforms

Engineering

Modelling & MDM

AI & analytics

Try it before you talk to us.

The same data, four layers up

Profile a CSV in seconds

Weekly demand forecast

Sample banking schema

Built by Informat.

RyuDB

Speed-up over DuckDB, TPC-H scale factor 10

Data and AI for the sectors that run Aotearoa.

The challenge

What we build

Example data product

The challenge

What we build

Example data product

The challenge

What we build

Example data product

The challenge

What we build

Example data product

Selected engagements.

Legacy analytics estate moved to Microsoft Fabric

Automated payments reconciliation

Industry reporting moved off legacy tools

Marketing data on autopilot

Senior people on every engagement.

Amir

ICT Security Specialist

Associate network

Tell us what's slowing your data down.

Before we start

Is your organisation ready for AI?

Start with a conversation.