The product
Two ways in.
One shared model.
DSharp models in both directions — top-down from a business question, bottom-up from the data you already have. It gets you to an agreed model fast, then turns it into governed, repeatable delivery. The result: one model your whole business can trust — sooner.
How it works
Work from either end
Top-down from a business need, or bottom-up from the data you already have — both roads land in the same model. Most teams use both.
Top-down
Start from the question
Begin with the question the business is asking. DSharp helps you model the concepts that answer it, then maps them to the data that fits.
- question
- →
- concepts
- →
- data
Bottom-up
Start from the data
Point DSharp at a source it has never seen. It profiles what's there, classifies it, and turns it into a model your team can agree on.
- source
- →
- profiled
- →
- model
Whichever way you start, you land on the same thing: one verified, reusable model — the shared definitions every report, dashboard, and AI query draws on. Data teams build it once; leadership reads the same answer everywhere.
Start from the question
Lead with the question, not the schema. You describe what the business needs to know; DSharp models the concepts behind it and maps them to the data that can answer — so the definition is agreed once, in business terms, and reused everywhere.
- Model the concepts a question needs — before touching a single table
- Agree definitions in business language, not column names
- Reuse the same concepts across every question that follows
Start from the data
Point DSharp at an unknown source and let it do the discovery. It profiles every table, proposes how they relate, and names and describes the concepts it finds — turning weeks of manual archaeology into a model you can review and refine. The source data itself is the hard part — this loop exists to get it right, not to pretend it's easy.
- Automatic profiling and classification of an unfamiliar source
- Proposed relationships you can accept, refine, split, or merge
- Business-readable names and descriptions, generated for you
- Catalogs document what exists — DSharp delivers what it means
- raw_0912
- tbl_appt
- sys_ref
- dbo.x_load
- z_tmp_44
One model, generated into delivery
Once the model is agreed, DSharp generates the structures and pipelines from it — organized into clear layers, delivered as standard code you own. No hand-built plumbing, no black box.
- Structures and pipelines generated automatically from the model
- Standard dbt you keep — inspectable, testable, repeatable
- Runs in your own environment, on the stack you already have
Generated, not hand-built
Runs dbt Core automatically in the background — significantly faster than doing it by hand.
Reusable by design
Every concept and its attributes are reusable — the more you model, the faster the next data product.
Technology-agnostic
The same model runs in the cloud or in your own data center — no vendor lock-in.
Built on the tools your team already uses
Start small
Prove it on one of your sources — in days.
No platform project. No big commitment. We take one real source — or just a model of it — and show you the speed and the trust live, in your own environment.