Data · Knowledge · Applied intelligence

coviny

Making knowledge work.

We connect evidence, decisions and action
to expand what organisations can do.

Independent thinking. Consequential engineering.

I. The practice

The world exceeds the models we make of it.

Data offers a partial account of a changing world. Patterns can reveal possibilities, but a correlation is neither a cause nor a guarantee. Useful intelligence makes its sources, assumptions and uncertainties open to examination.

Coviny brings data engineering, software development and applied artificial intelligence together to connect evidence with decisions and action. We build systems that can be questioned, revised and extended as conditions change.

II. Our disciplines

From evidence to action.

Data foundations

Connect sources, resolve entities and preserve the origin of each assertion. Distinguish observations from interpretations, and make gaps, conflicts and uncertainty visible.

Knowledge systems

Represent the objects, relationships and constraints that matter to an operation. Connect documents and records to a shared model that remains open to correction by new evidence.

Agentic workflows

Connect analysis to permitted actions in existing systems. Give agents defined tasks, explicit permissions and review points; record what was done and use the outcome to test the original assumptions.

Engineering & delivery

Focused teams shorten the distance between inquiry, implementation and use. Reusable components and automated testing turn each delivery into a foundation for the next.

III. Applications

Intelligence in use.

Start with a decision that matters, the evidence it requires and the actions it permits. These examples show how that connection can be built.

Agentic engineering

Carry a defined development task from issue to implementation, automated verification and a reviewable change. Our engineering harness coordinates isolated workspaces, bounded parallel execution, test and code-quality checks, and traceable run records. People define the task and review the resulting pull request.

Connected intelligence

Connect AI assistants to operational data through MCP and existing APIs. Authenticated access, structured queries and explicit system boundaries bring established records into the analytical workflow. Our existing implementation supports data access; transactional extensions are developed as separate, explicitly authorised capabilities.

Data enrichment

Reconcile fragmented sources into structured, usable datasets. Our work combines source registers, normalisation and field-level provenance with delivery manifests that make outputs and changes traceable. Existing market-specific scripts provide the foundation for a broader, reusable enrichment engine in development.

Research intelligence

Make research queryable. A structured model connects organisations, capabilities and relationships to their supporting evidence, confidence levels and unresolved questions. The implemented research database distinguishes proposed claims from validated findings, allowing comparisons and gaps to be examined on a common basis.

Quantitative intelligence

Make heterogeneous data comparable and calculations inspectable. Our work combines data normalisation, rule-based matching, duplicate resolution and indexed calculations across different formats and categories. Where direct observations are incomplete, explicit reference values and calculation rules support derived estimates. Measured values, assumptions and calculated results remain distinct, so the basis of a result can be examined and challenged.

Operational intelligence

Bring information from multiple sources into a shared operational model. Examine relationships, surface competing explanations and identify what further evidence is needed before acting.

Geospatial intelligence

Connect locations, postcode areas and geographic boundaries to a shared spatial model. Our engineering experience includes polygon-based coverage analysis, overlap-depth visualisation, GeoJSON map services and delegated access to location records. These foundations help teams understand where activities intersect and compare geographic coverage.

Tsutsumi — interoperable hand-offs

In development: a neutral adapter layer designed to translate a canonical selection into the receiving system’s required identifiers, semantics and format. Versioned mappings, rights checks, validation and explicit exceptions make differences visible rather than hiding them. The design includes a reviewable preview and traceable outputs; authorisation and execution remain with the receiving system.

Editorial intelligence

From source material to considered publication. AI-assisted drafting operates within a substantial, human-developed editorial framework: the principles and judgements through which sources are assessed, claims examined, arguments structured and meaning refined. Human editors direct the process, exercise critical judgement and retain responsibility for the final text and its publication.

Organisational knowledge

Connect approved documents and data to the decisions they inform. Retrieve answers with references, expose conflicting accounts and retain the distinction between established facts and provisional interpretations.

Intelligent commerce

Connect customer, product and transaction data to commercial decisions and authorised actions. Coordinate opportunities, offers and fulfilment across existing systems, with clear responsibilities and feedback from actual outcomes.

Illustrative applications; scope is defined for each engagement.

IV. Working method

Build capacity. Shorten the cycle.

Technology should expand an organisation’s capacity to understand and act. We look for bottlenecks in the passage from evidence to decision to execution, then redesign the process around a concrete, testable outcome.

Automate repeatable work, build reusable infrastructure and shorten feedback cycles. Speed is useful when it creates room for better judgement and further experimentation. Test consequences in use, revise assumptions and scale what holds up.

V. The team

The team

Twelve builders, researchers and shapers across Europe and Asia. Experience at Oracle, Adobe, SAP and Robert Bosch; academic backgrounds spanning Stanford, leading technical universities in Germany, Romania and Asia, and departments of philosophy and the social sciences. Several members hold multiple academic qualifications.

Deep data knowledge and AI fundamentals, grounded in mathematics, philosophy and social inquiry. We question assumptions, attend to detail and carry ideas into operation—even under difficult conditions and extreme pressure.

Our principles draw on Auftragstaktik, CDIO and shi: clear intent, independent judgement and sensitivity to circumstance. From conception to operation, thought remains accountable to execution.