Product
The evidence-backed context layer
Semantic Scout builds the context your agents and people work from — lineage, business meaning and business concepts — from how your data is actually made. It keeps that context current and attaches the evidence to every fact. No business data is touched.
Data lineage
Three views, one set of facts
All three are built from the same automatically harvested metadata, so they never drift apart the way hand-maintained diagrams do.
Business lineage
Sprawling technical flows summarised into clear, auditable stories a business reader can follow. This is the view a risk officer, an auditor, or a product owner actually reads.
Technical lineage
Full-fidelity, column-level lineage across jobs, SQL, and transformations — source to report, with change history for every hop. The view an engineer trusts for impact analysis.
System-level lineage
The estate at a glance: which systems feed which, where the mainframe boundary sits, and what a migration will actually touch.
The path a value takes
Real estates are full of decisions: the same field can end up in different places depending on what the code works out at run time. Tooling that reads structure alone can only report everything that might have contributed — which reads like an answer without being one.
Scout’s lineage reflects how the code actually decides, so you see the path a value takes and what part each input really played. Bring one program to a demo and compare it against what your current tooling tells you.
Data meaning
Meaning and business concepts
Catalogs learn what data means from how it is used — query history, dashboards, documentation. Scout derives it from what produces the number.
Understand your data
AI-generated descriptions, classifications including PII, and glossary links for every asset, derived from the code, configuration, copybooks, documents and logs behind it — reviewed and approved by your stewards, with version history behind each change.
Talk to your transformations
Ask what a pipeline, an SQL job, or a COBOL program does and get a plain-language answer grounded in the actual code. Understanding that used to require the original author — including for programs whose authors retired years ago.
Business concepts
Concepts hold the meaning; terms are just the words for them — so a concept keeps its meaning across teams and systems. Each is bound to the assets that produce it, and business lineage follows business variables and business steps, not only tables.
Chat over COBOL is part of what a Nordic bank licensed us for — see context for modernisation.
Quality
Evidence, and the gaps have names
An answer with a file and a line behind it survives an audit. A confident answer with nothing behind it does not.
Accurate
Claim correctness, measured on development cases and on cases withheld from development. A wrong attribution counts as a failure, not a near miss.
Complete
Coverage and closure, with gaps attributed. Where something is expected but uninvestigated, or outside the scope you granted, it is named as such.
Current
Checked against the source revision on every read, and refreshed when a new snapshot arrives.
Evidence-backed
Every fact cites the file and line it came from. Open the evidence straight from the lineage.
Readiness
Context that knows when it is stale
Context decays the moment nobody is watching it. Scout watches — and says what it does not yet know.
Current on every read
Every read checks currentness against the source revision. When code changes, only the meaning that depends on it is re-evaluated — not the whole estate.
Asks instead of guessing
When an agent needs context that is missing or stale, it can request it, rather than reasoning from whatever it happens to find.
Disagreement stays visible
When your catalog, your code and an expert disagree, Scout returns one explainable answer and keeps the disagreement inspectable, rather than letting the last write win.
Sources and consumers
Reads what your systems leave behind
Semantic Scout works from what you already have rather than asking you to move to something new — and one graph serves everyone who needs it.
Code and configuration
- Code repositories
- Configuration and copybooks
- Common SQL dialects, Python, Spark
- dbt and orchestration tools
Documents and logs
- Specifications and design documents
- Operational logs
Mainframe
- COBOL programs
- Mainframe datasets and jobs
Platforms
- Cloud warehouses and lakes
- BI tools
Niche or in-house system? Ask us about your stack — we build the connector.
Who uses the result
Your people
Explore the graph, read the business view, and open the evidence behind any fact.
Your AI agents
Query the same lineage and meaning directly over MCP and A2A, with the evidence attached, so an agent’s answer can be traced like a person’s.
Your catalog
Kept in sync, so Collibra, Ab Initio, IBM or DataHub stays the front door for everyone who already works there.
Deployment & security
Packaged and deployed to fit your constraints
Plug-in
The Semantic Scout Harvester and UI running alongside the catalog and quality tools you already own.
Enterprise bundle
Catalog, quality, and lineage delivered together and pre-wired — DataHub and Great Expectations included.
Deploy self-managed, as dedicated SaaS, in your VPC, or fully on-premise — chosen by your security and data-residency needs.
- No business data is used for metadata generation — none.
- Runs against configuration, logs, and repositories through APIs, and can be isolated from production workloads.
- In-region deployment keeps metadata and logs inside your chosen geography, including full EU residency.
Bring a repository. Leave with its context.
That is the fastest way to find out whether we are right about your estate.