Data & AI platform
One definition of revenue.
Every dashboard obeys it.
Atmora Lens keeps metric definitions in a version-controlled semantic model that every chart, scheduled report, embedded widget and natural-language answer resolves through. The month-end argument moves from whose number is right to what the number means.
- Cloud
- Private cloud
- On-premise
Overview
A governed semantic layer with dashboards attached, so every report agrees on the numbers.
Atmora Lens is a semantic layer with dashboards attached. Active customer, recognised revenue, contribution margin and the other definitions that matter are declared once in a model reviewed through pull requests, and everything downstream resolves through them. Queries push down to your warehouse, with incremental materialisation chosen from observed query patterns rather than from an engineer guessing which aggregate table to build next. Lineage runs from any figure on screen back through the model to the warehouse tables and dbt models behind it.
Lens assumes you already have a warehouse or lakehouse worth querying. It is not a data pipeline, it will not replace ingestion or transformation tooling, and it reads dbt metadata rather than duplicating it. If your reporting is three spreadsheets and a monthly board pack, this is more governance than the problem deserves. It becomes worth the setup effort when several teams publish numbers, those numbers disagree, and reconciling them costs more each month than the tool does.
- 3.4 s
- P95 dashboard load over a 2.1 billion-row fact table
- -58%
- Warehouse compute spend after incremental materialisation
- 340
- Dashboards resolving through one revenue definition at Halden Group
- 9
- Working days from first model commit to a governed board
Modules
What ships in the box
- Semantic model and metric registry
- Query engine, cache and materialisation
- Dashboard and report builder
- Row and column-level governance
- Scheduled distribution and metric alerts
- Embedded analytics and natural-language query
Integrates with
- Snowflake
- Databricks
- BigQuery
- PostgreSQL
- dbt
- Salesforce
- SAP BW
- Microsoft Excel
Capabilities
What it does
Metrics defined once, in version control
Revenue, churn and margin live in a semantic model reviewed through pull requests. Charts reference the definition, so nobody quietly redefines margin inside a dashboard filter.
Materialisation driven by real queries
The engine decides what to pre-compute from observed query patterns, refreshes only partitions that changed, and falls back to live SQL whenever a request misses the cache.
Governance on the model, not the dashboard
Access rules attach to entities and fields. A regional manager opening a shared board sees only their rows, and restricted columns are absent from the result rather than masked on screen.
Lineage from a chart back to the source table
Every figure traces through the model to warehouse tables and dbt models. When an upstream job fails, the dashboards it feeds are flagged before somebody screenshots one for a board pack.
Natural-language query that shows its SQL
Ask in plain English and get the generated SQL next to the answer. Queries are confined to the semantic model, so the assistant cannot invent a join the business does not recognise.
Embedded analytics with tenant-aware tokens
Governed charts drop into your own product behind signed tokens carrying tenant and role. Theming follows your CSS variables instead of an iframe that looks borrowed from elsewhere.
Distribution people actually open
Boards go out as PDF, XLSX or a chat summary on a schedule or a data-arrival trigger, with each subscriber's row filters applied before the file is generated and sent.
Drift alerts on metrics, not just charts
Set an expected range, a week-on-week band or a freshness window on a metric and get told when reality leaves it, with the contributing dimensions ranked by impact.
Built for
Where this is deployed
Industry
Banking & Financial Services
Payment rails, risk engines and core-adjacent systems built by people who have run them at 03:00 on a settlement night.
Industry
Retail & E-commerce
Storefronts, stock ledgers and fulfilment logic that hold together on your busiest trading day.
Industry
Manufacturing
Getting the plant floor and the ERP to agree on what actually happened during the shift.
Industry
Logistics & Supply Chain
Visibility, customs filing and cost-to-serve systems for operators who move real freight.
Industry
Energy & Utilities
Meter data, settlement and grid integration for organisations where a rounding error becomes a regulatory finding.
Industry
Healthcare & Life Sciences
Clinical and regulated software built to pass an audit, not just a demo.
FAQ
Questions we get asked
- Does our data leave the warehouse?
Only the result sets needed to draw a chart, and only into the cache tier you configure. Materialised aggregates can be written back into your own warehouse schema instead of ours, which is how most banking and healthcare deployments are set up.
- We already use dbt. Is the semantic model a second copy?
No. Lens reads dbt models, tests and lineage directly and layers metric definitions on top of them. Where dbt already defines a metric, Lens imports it rather than asking you to restate the same logic in another language.
- What does it not do?
It does not ingest or transform data, does not schedule ELT, and does not try to be a notebook environment for data scientists. Lens governs and serves numbers; the pipeline that produces them stays with your existing tooling.