Data Lab.One version of the numbers, ready to be questioned.
We build warehouses, pipelines, semantic layers and dashboards, then put governed talk-to-your-data agents on top. Metric definitions are agreed and tested, so the board pack and the dashboard say the same thing.
Build · Run · 24/7
01What we build
What the Data Lab builds.
Archetypes, named by what they do. Every engagement also carries the standards we apply in every lab.
- 01
Warehouses and lakehouses
Snowflake, BigQuery or Postgres, modelled with dbt around the questions the business actually asks.
- 02
ELT pipelines
Ingestion from SaaS tools, databases and files: scheduled, tested and monitored.
- 03
Semantic layers
Metrics defined once and reused by every dashboard, notebook and agent.
- 04
BI dashboards
Dashboards built for a decision, with an owner, a definition for every number and a refresh you can trust.
- 05
Talk-to-your-data agents
Plain-English questions answered from governed models, with the query shown and permissions respected.
- 06
Customer-facing data products
Usage analytics, reporting and exports embedded in your own product.
02Services
Services and deliverables.
Concrete scopes with named outputs, so you know what you will hold at the end of each stage.
Data platform build
Warehouse, ingestion and modelling set up as code, with tests and lineage from the first model.
- Warehouse and dbt project
- Ingestion pipelines
- Lineage and documentation
Analytics and BI
Dashboards and reports with agreed definitions and named owners.
- Metric catalogue
- Dashboards
- Owner training
Governed AI on data
Talk-to-your-data agents over the semantic layer, using the dbt, Snowflake, BigQuery or Postgres MCP servers, with access control and query logs.
- Data agent
- Access policies
- Query audit log
Data operations
Monitoring, quality alerts and cost control as a retained service.
- Quality monitors
- FinOps report
- Incident handling
03Live tool: CSV to dashboard
Drop in a CSV.
Your file is processed entirely in this browser tab. You get a column profile, data quality flags and first charts in seconds.
04Lab standard
The A* standard for this lab.
On top of fixed-price discovery, a named lead, code you own, test gates, observability, written decisions and an SLA-backed run retainer, this lab adds:
- 01
A semantic layer
Revenue, churn and margin are defined once, in code, and every tool uses the same definition.
- 02
Data contracts
Producers and consumers agree schemas and expectations, so breaking changes fail before they reach a dashboard.
- 03
Quality tests and lineage
Freshness, volume and validity are tested on every run, and every number can be traced to its source.
- 04
FinOps for data
Warehouse spend is monitored by model and team, with alerts before a query surprises the budget.
- 05
Governed agents
Data agents answer only from approved models, within the asking user's permissions, and show their working.
05Human sign-off
Where people sign off.
Pipelines and agents can be automated. Ownership of data, and of what the numbers mean, belongs to named people in your organisation.
- Data ownershipEvery dataset has a named owner on your side who decides who may use it and how.
- Lawful basis under UK GDPRYour data protection lead confirms the lawful basis for each use, and pipelines are designed to match it.
- Metric definitionsWhat counts as an active customer or recognised revenue is agreed by your finance and operations leads, then encoded.
06FAQ
Questions, answered.
Which warehouse should we use?
It depends on data volume, your existing cloud, team skills and cost profile. Postgres is often the right start, with Snowflake or BigQuery when scale or concurrency demands it.
Does the CSV tool upload our data?
No. The demo on this page processes the file in your browser and sends nothing to a server.
Can you work with our existing Power BI or Looker estate?
Yes. We often keep the BI tool and fix what sits underneath it: models, definitions and data quality.
How do talk-to-your-data agents avoid wrong answers?
They query a governed semantic layer instead of raw tables, show the query they ran, and decline questions outside the approved models.
08Contact
Bring us the numbers nobody agrees on.
Your first call is with an engineer from the Data Lab, not a sales script. Bring the system, the deadline and the constraints.
hello@aurionlabs.io