Capability map
Three pillars, more than sixty individual capabilities.
We do not sell packages, we do concrete work. Below is the full list of what we do: for every capability you will find what it means in practice and what we actually hand over.
Governance
Trust in the data
Data can support decisions only when you know who is responsible for it, what it means, how reliable it is, where it comes from, and who can access it. This pillar builds that foundation of trust: not with policy documents, but with working practice and controls.
Who owns the data
accountability
Appointing data ownersdata ownership
What it means
Every important data domain, such as customer, contract or transaction, has a named business owner who decides on its definition and its access.
What we hand over
A data owner map: which data domain belongs to whom, with what decision rights, and how stand-in arrangements work.
A network of data stewardsdata stewardship
What it means
Colleagues working alongside the data owner who handle the day-to-day tasks: refining definitions, flagging errors, watching quality.
What we hand over
Named stewards, with role descriptions, time-commitment estimates and onboarding training.
Decision forum and dispute resolutiongovernance forum
What it means
A regular forum where contested questions of meaning and access get decided, and where issues left unresolved locally are escalated.
What we hand over
The forum's charter, its membership, its meeting cadence, and a description of how an undecided question moves up.
Onboarding new dataintake process
What it means
A defined procedure for the conditions under which new data or a new source system may enter the company's data assets.
What we hand over
An intake checklist and approval workflow, built into your existing change management.
What the concepts mean
meaning and semantics
Business glossarybusiness glossary
What it means
Approved definitions of the company's key concepts, kept in a single place: what an active customer, a closed contract or revenue means.
What we hand over
A populated business glossary covering the most contested concepts, with an approval loop and a maintenance routine.
Calculation rules for metricsmetric definitions
What it means
The metric's name is not enough: the formula, the filters, the time period and the exceptions must be written down as well.
What we hand over
A metrics handbook for every number in the executive reports, with unambiguous, reproducible calculation descriptions.
Unifying code listsreference data
What it means
A consistent, maintained meaning for the codes used across systems, such as statuses, product types and country codes.
What we hand over
A central code dictionary for the important code lists, with cross-system mapping.
Resolving divergent terminologysemantic reconciliation
What it means
Systems often mean different things by the same name. These differences must be uncovered and consciously resolved.
What we hand over
A divergence map for the main concepts, with a decision on which meaning is authoritative and how the others map to it.
Keeping shared concepts alignedfederated governance
What it means
As more teams start using the same concepts, such as customer, product and contract, you need a discipline that keeps their meanings from drifting apart.
What we hand over
A versioned shared concept set with a named owner, a review routine and a decision log.
Can the data be trusted
data quality
Defining quality expectationsquality rules
What it means
Measurable rules for when data is good: how fresh it is, how complete it is, whether it is duplicated, whether the value is valid.
What we hand over
A rule set per data domain, phrased in business language, then translated into machine-executable checks.
Automated checks and alertingquality monitoring
What it means
The rules do not live on paper: they run on every load, and any deviation triggers an alert to the responsible person.
What we hand over
Checks wired into the critical data flows, with an alerting channel and assigned owners.
Issue handling processissue management
What it means
When bad data surfaces, it needs a clear path: who logs it, who fixes it, and what happens to the consumers in the meantime.
What we hand over
An issue handling procedure, with a register and root-cause analysis of recurring errors.
Data contractsdata contracts, ODCS
What it means
A written, machine-readable agreement between the team providing the data and the team consuming it: what the data contains, at what quality, and by when it is available.
What we hand over
Data contracts for the most important data hand-offs, with automated validation in the loading pipeline.
Comparing documented behavior with realityas-is analysis
What it means
We compare how the company is supposed to work, according to policies and system documentation, with what the data actually shows. The gap is not an error but valuable information about how the company really operates.
What we hand over
A gap list between documented and actual behavior, marking which findings the real data has already confirmed.
The same customer in every systementity resolution
What it means
The same customer or product appears in several systems under different identifiers. Unless this is reconciled, every report counts something different.
What we hand over
Unified identification rules and cross-system matching: the most commonly underestimated work in any rollout; we plan for it from day one.
Where it comes from and where it goes
traceability
Mapping data lineagedata lineage
What it means
You can trace how a number seen in a report came to be: from which source system, through which transformations.
What we hand over
A lineage map for the critical metrics, displayed in the catalog and refreshed automatically.
Impact analysis of changesimpact analysis
What it means
Before a source system changes, you need to see which reports and processes are affected.
What we hand over
An impact view that shows what a planned change touches.
Data catalogdata catalog
What it means
A searchable register of what data exists in the company, who owns it, and what it means.
What we hand over
A populated catalog of the important data assets, with descriptions, owners and a search interface.
Who can access it and how long we keep it
security and compliance
Classifying data by sensitivitydata classification
What it means
Every data domain gets a classification: public, internal, confidential or special-category personal data.
What we hand over
A classification scheme and the actual classification of the important data domains, together with the protection requirements.
Access modelaccess control
What it means
Who may see what, on what grounds, and how access is requested or revoked.
What we hand over
A role-based access model, with request and review workflows.
Retention and deletionretention
What it means
Which data must be kept for how long, when it must be archived, and when deletion is mandatory.
What we hand over
A retention rulebook per data domain, with automated enforcement.
Handling personal dataprivacy, GDPR
What it means
The legal basis and purpose limitation of personal data processing, and serving data subject rights.
What we hand over
A record of processing activities and the practical fulfillment workflows, such as data portability.
Is AI use governed
AI governance
AI inventoryAI inventory
What it means
A register of the artificial intelligence solutions running in the company, including vendor-supplied and embedded features.
What we hand over
A complete AI inventory by system, with an owner, the data domains used and the business purpose.
Risk classificationrisk classification
What it means
Every AI solution must be classified by risk, because the obligations follow from it.
What we hand over
Classification following the logic of the EU AI Act, with the high-risk systems flagged.
Control points and human approvalAI controls
What it means
Written rules on where human approval is mandatory, what must be logged, and how model performance is measured.
What we hand over
A single control framework that covers the expectations of NIST, ISO 42001 and the EU AI Act in one system.
Logging AI useauditability
What it means
It can be proven after the fact which data, which model and what moment an AI decision was based on.
What we hand over
A logging regime and the technical foundations of provability.
Enabling
Capability in the organization
Technology on its own does not deliver results: most rollouts stall for organizational reasons, not technical ones. This pillar builds up your own people, so they understand, own and can carry the data work forward after our involvement ends.
Where data creates value
business direction
Discovering value opportunitiesuse case discovery
What it means
In executive workshops we go through where better data would bring measurable benefit: where money, time or customers are being lost.
What we hand over
A ranked list of opportunities, with estimated business impact and implementation difficulty.
From business need to data requirementrequirements engineering
What it means
We turn a stated business need into concrete data requirements: what data is needed, at what quality, how fresh.
What we hand over
A requirements catalog that can be handed straight to development.
Measuring the returnvalue tracking
What it means
We agree up front how success will be measured, and after go-live we actually measure it.
What we hand over
A measurement frame for the selected opportunities, with baseline values and post-implementation review.
Does the organization understand data
knowledge and capability
Executive data awarenessexecutive literacy
What it means
Senior leadership gets a realistic picture of what data and AI can do, and what they cannot.
What we hand over
An executive program series, with decision-support briefs, built on your own company's examples.
Data literacy for staffdata literacy
What it means
Business colleagues learn to ask, interpret and argue with data, through examples from their own work.
What we hand over
A tiered training program, with hands-on exercises and measurable outcome requirements.
Training data stewardsrole-based training
What it means
The appointed stewards learn what their role means in practice and which tools they work with.
What we hand over
Role-specific training and mentoring on live tasks, not sample data.
Preparing internal trainerstrain the trainer
What it means
The company's own people become able to train their colleagues further.
What we hand over
Trainer preparation, with handed-over course materials and a practice environment.
The company's own vocabularycanonical vocabulary
What it means
The concept set is built in the company's own language, not in the system vendors' vocabulary; vendor terms are attached to it as mappings.
What we hand over
A concept set owned by the company and a mapping table to the vendor dictionaries. Interpretation does not stay locked inside a supplier's vocabulary.
Joint concept-building workshopsworkshops
What it means
We do not invent the concepts and rules for you: we uncover them with your experts in workshops, starting from an industry reference model, not from a blank page.
What we hand over
A leadership overview session, then a discovery series with documented results. The model is the product of joint work, so the organization treats it as its own.
How the data organization should work
operating model
The data organization's operating modeltarget operating model
What it means
We design the roles, processes and forums needed so that data work does not depend on a few enthusiastic people.
What we hand over
An operating model design with roles, responsibilities and a rollout timeline.
Center of excellencecenter of excellence
What it means
A central team that supports the business areas as a service, while expertise also grows locally.
What we hand over
A setup plan for the center of excellence: headcount, service catalog, collaboration routines.
Appointing data product ownersdata product owner
What it means
A responsible person in the business area who runs their data product like a product: with consumers, a roadmap and quality.
What we hand over
Trained owners who carry their own area's data product forward independently.
Does it stick in daily work
rollout and adoption
Adoption planadoption
What it means
We think through where the new solution appears in whose daily work, and what it takes for people to actually use it.
What we hand over
A role-by-role adoption plan with the integration points and the support needed.
Measuring usageusage analytics
What it means
We do not assume usage, we measure it: who uses what, and how often.
What we hand over
Usage metrics and regular reviews, with intervention proposals.
Playbooks and templatesplaybooks
What it means
We turn the work done together into documented, repeatable playbooks.
What we hand over
Playbooks with templates, for example for defining a data product or introducing a contract, with a handover session.
Independent expert opinionadvisory
What it means
We evaluate technology directions and incoming proposals free of vendor interest.
What we hand over
Decision-support material with a clear recommendation and the risks named.
Progress tied to acceptance criteriagated delivery
What it means
Every step has measurable acceptance criteria agreed in advance: for example, that the system answers 10–20 of your real business questions on real data. Until a step passes its own gate, we do not move on.
What we hand over
A gate list for the rollout steps, with measurable criteria and joint evaluation.
Platform
Open architecture
Data has to travel from source to use, reliably and affordably. We design this layer on open standards, so your data assets are not tied to a single vendor and the technology decision remains yours later on.
How the data gets there
data movement
Change-based data transferCDC
What it means
We move only the actual changes out of the source system, continuously, without putting load on it.
What we hand over
Change capture wired into the critical sources, with an agreed freshness commitment.
Event stream processingevent streaming
What it means
Where the business need calls for it, we process data as it arrives, not batched by the hour.
What we hand over
Operable, monitored event streams, proven in production on live banking streaming platforms.
Scheduled bulk loadingbatch ELT
What it means
Where real time is not justified, we build scheduled, controlled batch loading.
What we hand over
Scheduled loading pipelines with error handling and re-runnability.
Feeding results back into operational systemsreverse ETL
What it means
Analytical results go back to where the work happens: into the CRM, onto the agent's screen.
What we hand over
Feedback data flows into the selected operational systems.
Where we store it
storage and format
Layered data storagelakehouse, medallion
What it means
Data lives side by side in raw, cleansed and business layers, so every use works from the right level.
What we hand over
A layered storage architecture, with rules and responsibilities between the layers.
Open table formatopen table format
What it means
The data sits in a format that several vendors' tools can read, so it is not tied to a single platform.
What we hand over
Open-table-format storage, connected to and tested with multiple engines.
Open catalog layerREST catalog
What it means
The register of stored data is decoupled from the processing engine, so the engine can be swapped at any time.
What we hand over
A vendor-independent catalog, connected to multiple processing tools.
Preserving the option to switch vendorsexit strategy
What it means
We think through in advance what happens if you have to switch: at what cost and in how much time.
What we hand over
A dependency assessment and a documented exit scenario.
How we transform it
processing
In-warehouse transformationtransformation as code
What it means
The steps that turn raw data into business data are written as version-controlled, tested code, not as manual queries.
What we hand over
A version-controlled transformation codebase with automated tests and documentation.
Large-scale computationdistributed processing
What it means
Where large volumes of data must be processed at once, we use a distributed computing framework.
What we hand over
Optimized processing jobs, with runtime and cost measurement.
Real-time computationstream processing
What it means
Immediate aggregation, filtering or alerting on continuously arriving events.
What we hand over
Event-driven processing logic, with operations documentation.
Scheduling data pipelinesorchestration
What it means
Reliable scheduling of multi-step data pipelines, with dependency handling and re-runs.
What we hand over
An orchestration layer in which one failed step does not take down the whole overnight run.
How users get access
access and reporting
A single layer of metrics and conceptssemantic layer
What it means
Metrics and dimensions defined in one place, which reports and AI assistants both connect to. If two screens were to show different numbers, the rule is clear: the shared semantic layer is the source of truth.
What we hand over
A production semantic model with the key metrics, including permission handling.
One query across many sourcesdata virtualization
What it means
Scattered systems can be queried through a single interface, without moving the data.
What we hand over
A federated query layer over the selected sources, with access control.
Natural language queryingnatural language query
What it means
Business users ask in their own words and get reliable answers backed by the semantic layer.
What we hand over
A production natural language interface for one business area, with measured accuracy.
Self-service reportingself-service BI
What it means
The business areas build their own reports, but from shared, authoritative definitions.
What we hand over
A reporting framework and the training that goes with it, so self-service does not descend into chaos.
How we run AI
AI capabilities
The full model lifecycleML platform
What it means
The discipline for developing, versioning, deploying and monitoring machine learning models.
What we hand over
A working model lifecycle, wired up and monitored with the first models.
A shared foundation for language modelsLLM platform, RAG
What it means
A common foundation that several AI applications can build on: document processing, retrieval, context handling.
What we hand over
A central LLM foundation that applications share in a controlled way.
Governed data access for AI agentsgoverned agents
What it means
AI agents do not query raw databases: they work through the semantic layer, logged and permission-controlled.
What we hand over
A governed access layer for the agents, with comparative accuracy measurement.
How we operate it
operations
Version-controlled infrastructureinfrastructure as code
What it means
The platform's configuration lives in code, so changes are traceable and reversible.
What we hand over
Infrastructure described in code, with automated deployment.
Automated testing and releaseCI/CD
What it means
Changes to data pipelines go live through the same controlled path as software changes.
What we hand over
A testing and release process for the data platform.
Monitoring and alertingobservability
What it means
We see when a data flow stops, slows down or produces odd values, before the business notices.
What we hand over
A monitoring layer over the critical pipelines, with alerting and on-call routines.
Cost transparencyFinOps
What it means
Spend on the cloud data platform can be tracked by business area and by pipeline.
What we hand over
Regular cost reporting and optimization proposals.
All 66 capabilities point in one direction: your data assets in your organization's hands, with the knowledge staying in-house.
In our work we align with DAMA-DMBOK, EDM Council DCAM, the ODCS data contract standard, NIST AI RMF, ISO/IEC 42001 and the EU AI Act. But we do not sell frameworks, we deliver implementation.
Start with a data asset audit