Language and tooling
Make system knowledge usable
Explore the building blocks for describing domain rules, coordinating work, and connecting system definitions to running software.
About Cohesive Systems
Understand your software.
Change it with intent.
Cohesive gives teams and AI agents a shared, reviewable account of how a software system works. Our language and compiler tooling connect product decisions and technical constraints to code, infrastructure, and runtime evidence, so each change can build on what the team already knows.
What Cohesive does
Cohesive Systems builds a language and toolchain for describing software systems. Teams express the rules of their domain, the actions a system allows, and the requirements it must meet as readable, versioned definitions. These definitions form a semantic system graph that connects accepted decisions to implementation.
That gives developers and agents a place to find the meaning behind the code: what an entity represents, which changes are valid, and which obligations a workflow carries across services.
Product
Connect data models with Ari
Ari uses semantic and statistical inference to propose relationships between data models, with evidence that people can review.
What makes Cohesive different
System meaning often lives across code, documents, schemas, and team memory. Cohesive gives that meaning an explicit form that tooling can inspect and use throughout a system's life.
Decisions become part of the system
Domain rules, policy, authority, and operational requirements have readable definitions that teams can version and review as the system changes.
Meaning stays connected to implementation
Typed mappings connect definitions to code, APIs, workflows, infrastructure, tests, and runtime evidence. Teams can examine how an implementation carries out an accepted decision.
People and agents share context
The graph supplies context for development agents and declares the actions, processes, and authority that guide agents interacting with a running system.
Who uses Cohesive
Cohesive is built for people responsible for understanding and evolving software systems. Its intended users include:
- Engineering teams and architects working through domain models, service boundaries, distributed workflows, and changes that span multiple codebases.
- Teams building with AI agents who need accepted rules and constraints to remain available as agents develop or operate software.
- Data and integration teams connecting models across systems, where the meaning of a field or relationship matters as much as its shape.
- Founders and engineering leaders deciding how an early product can grow into a durable system.
A starting point can be one entity, relation, process, or integration. Existing services and databases continue to do their work as their meaning becomes easier to inspect.
The team behind Cohesive
Cohesive Systems was founded by Leo Gorodinski after more than 25 years building and scaling B2C and B2B systems. Across product, engineering, and operations, he kept encountering the same problem: important system meaning was scattered across code, tools, and team memory. Cohesive grew from that experience.
The work brings language design, compiler tooling, and product development together, testing the ideas against concrete implementation problems.
Cohesive welcomes technical contributions and research, product, and delivery partnerships. Get in touch to discuss a system you are working on or a contribution you would like to make.
How Cohesive works
The process begins with a bounded part of a system and the decisions that govern it. Those decisions become source that can be reviewed and compiled.
- Describe the intent. Define entities, relationships, valid transitions, policy, and operational requirements in readable semantic source.
- Compile the graph. Resolve shared identities and dependencies into graph state, with semantic diffs and diagnostics that help teams review changes.
- Connect the implementation. Use typed mappings to connect definitions to code, schemas, APIs, workflows, infrastructure, and tests.
- Use evidence to guide change. Bring runtime evidence and implementation context back into the next decision, so people and agents can work from the system's accumulated knowledge.
Explore the building blocks for the technical details, or read how Cohesive supports AI agents.
Key facts
Company
- Detail
- Cohesive Systems
Founder
- Detail
- Leo Gorodinski
Founder's experience
- Detail
- More than 25 years building and scaling B2C and B2B systems
Focus
- Detail
- Semantic system graphs, language design, and compiler tooling
Product
- Detail
- Ari: semantic and statistical inference for connecting data models
Systems work
- Detail
- Selective engagements in architecture, domain modeling, distributed workflows, and technical strategy
Contact
- Detail
- hello@cohesivesystems.com
Frequently asked questions
What is a semantic system graph?
It is a connected representation of a software system's meaning: its domain concepts, rules, structure, operational requirements, and the mappings to what runs. Cohesive compiles readable definitions into this graph so the relationships can be inspected and used by tooling.
How does Ari relate to Cohesive?
Ari is a Cohesive product focused on connecting data models. It uses semantic and statistical inference to propose relations and gives people evidence to review those proposals. The broader Cohesive language describes system meaning beyond data relationships, including behavior and operational requirements.
Do we need to replace our existing systems?
Adoption can start with a single entity, relation, process, or integration. Cohesive connects definitions to existing databases, services, workflows, and interfaces, allowing teams to introduce it within a bounded area.
How does Cohesive support AI agents?
Development agents use the graph to understand a system and the constraints on a change. Agents interacting with the running system use its declared actions, processes, policy, and authority. Both draw on accepted intent and evidence.
Can we work with Cohesive on an architecture problem?
Cohesive Systems works selectively with founders and engineering leaders on architecture, domain modeling, distributed workflows, and technical strategy. An engagement can begin with the problem you are facing, including work that does not begin as a Cohesive implementation.
How can I get started or contribute?
Contact us to request a demo or discuss a specific system. For technical exploration, start with the building blocks and the Cohesive repository, where you can explore the code, open an issue, or discuss a contribution.