Architecture knowledge is fragmented, stale, and scattered across people and systems. Kyb continuously infers your organisational architecture from evidence, not manual modelling. One semantic graph. Every system, process, and dependency. Queryable, governed, alive.

Open standards and semantic rules, not a proprietary black box.
Your models, your keys — no vendor lock-in on the intelligence.
Nothing enters your graph until you accept, reject, or reclassify it.
Your architecture stays yours. We don't train on or sell it.
Architecture repositories go stale because humans must manually maintain them. Critical knowledge is scattered across engineers, wikis, cloud consoles, and tribal memory. Traditional tools are expensive and time-consuming, built around manual modelling of what everyone already half-knows.
Kyb infers your architecture from fragmented evidence and refines it through targeted conversation. The AI classifies systems, proposes relationships, and detects contradictions. You correct what it gets wrong. The graph converges toward reality, not away from it. Every fact carries confidence, provenance, and history.
Kyb infers structure from real evidence and refines it through conversation — so your model stays current without anyone manually re-drawing it.
ArchiMate constraints and semantic rules surface gaps, invalid dependencies, and drift automatically — before the steering committee finds them.
Enterprise, solution, data, and business views are all projections of a single canonical graph — not separate repositories that drift apart.
Every fact carries a confidence score, source attribution, and history — so you know exactly what's inferred, what's confirmed, and what's gone stale.
When the AI doesn't know something, it asks the right person the right question — maximum graph improvement per human interruption.
Import existing Archi models as a starting point. Kyb fills the gaps, detects what's changed, and keeps the model alive as your organisation evolves.
Point Kyb at your GCP project and it auto-discovers VMs, GKE clusters and workloads, Cloud Run services, networks, load balancers, Pub/Sub, Cloud Functions, and 30+ resource types. The AI maps raw infrastructure into architecture-level abstractions — application components, technology services, and their relationships — so you get a living architecture model, not a cloud inventory.
Review every proposed element before it enters your graph. Accept, reject, or reclassify. Your architecture, your judgement — the AI does the heavy lifting.
VMs, GKE, Cloud Run, networks, Pub/Sub, and 30+ resource types
Raw infrastructure mapped to ArchiMate application and technology layers
Review and approve every proposed element before it enters the graph
AWS, Azure and Cloudflare discovery on the roadmap

Kyb does not rely on a rigid database schema. The entire platform is powered by a dynamic ontology built on W3C semantic web standards (RDF, OWL, and SHACL). This enables a fully data-driven, user-customizable metamodel that conforms to standard enterprise frameworks while adapting seamlessly to your organization's unique vocabulary.
Define your own enterprise, technology, and business objects, connect them with customizable relationship structures, and query the entire landscape instantly.
Every component, process, and connection is stored in a standardized, machine-readable semantic web format for absolute interoperability
Define custom validation shapes to enforce architectural standards (e.g. 'Database components must belong to a Virtual Machine')
Write declarative queries to extract complex multi-hop dependencies and generate real-time reports without writing custom code

Free tier. Paid plans coming soon.
For a single project.
For solo architects.
For architecture teams.
Free tier. No credit card. One architect, one afternoon, full visibility.
Try the beta