MUKR
The knowledge layer that captures why decisions are made — documents, wikis, notebooks, and 3D annotations with evidence, confidence, and provenance built in.
MUKR
The knowledge layer that captures why decisions are made — documents, wikis, notebooks, and 3D annotations with evidence, confidence, and provenance built in.
Evidence, Not Just Documents
Every claim carries its evidence chain, confidence score, provenance graph, and review state. When someone asks 'why do we believe this?', the answer is one query away.
Real-Time Collaboration
CRDT-native editing means two engineers write the same paragraph simultaneously — both edits survive. Offline first. Conflict-free merge on reconnect. No locking.
Universal Cross-References
Type @ and reference any entity across your operations platform, object platform, datasets, models, tickets, and documents. Backlinks update automatically.
Computational Notebooks
Run Python, Rust, SQL, and Markdown in reactive cells next to institutional knowledge. Reproducible analysis with evidence attachment and provenance.
3D Annotations
Open GLTF, OBJ, STL, STEP, and IGES models in the browser. Annotate parts, take measurements, cross-section, and overlay analytics — all tied to the knowledge graph.
Knowledge Portals
Publish curated knowledge into branded customer portals via the experience platform. 35+ component types for graph viewers, entity cards, wiki embeds, and 3D viewers.
Every organization loses knowledge. The engineer who designed a part leaves. The analyst who wrote the risk assessment moves teams. The compliance rationale behind a policy decision gets buried in email. The data remains — in ERP, PLM, CRM, ticketing systems — but the why disappears. MUKR captures the why.
MUKR is a knowledge context layer built in Rust. It combines documents, wikis, computational notebooks, 3D annotations, and a knowledge graph into one platform where every claim carries its evidence, its confidence, its provenance, and its review state. When a design decision is documented, the rationale, the alternatives considered, the evidence supporting it, and the reviewers who approved it travel together — not as scattered metadata, but as structured knowledge.
Knowledge Surfaces
Most knowledge tools are one thing — a wiki, a document editor, a notebook, a 3D viewer. Teams stitch them together with links that break, screenshots that go stale, and references that nobody can trace. MUKR puts every knowledge surface on one platform, connected by the same knowledge graph, the same collaboration engine, and the same evidence system.
A block-based document editor with 35+ block types — paragraphs, headings, lists, media, tables, code, formulas, notebook cells, 3D models, cross-references, and knowledge claims. Drag and drop reordering. Schema-validated blocks so structured content stays structured.
Every block is CRDT-native — two people editing the same paragraph is a normal operation, not a conflict. AI-assisted composition for summarize, expand, improve, translate, and explain. When a block references an entity in your operations platform, the reference is live — change the entity and the document reflects it.
How Claims Become Knowledge
A wiki stores pages. A document store stores files. Neither tells you whether the information is trusted, who reviewed it, what evidence supports it, or whether it has been superseded. MUKR treats knowledge as an evidence system — every claim has a lifecycle, a confidence score, a provenance graph, and a review state. Knowledge is not what is written. Knowledge is what has been evidenced, reviewed, and accepted.
A claim moves through 11 states: Proposed, Evidence-Attached, Under-Review, Validated, Verified, Operationalized, Disputed, Stale, Superseded, Retracted, and Rejected. Each transition requires provenance, calibration, policy approval, and a recorded owner. A claim does not become operational knowledge because someone wrote it — it becomes operational knowledge because it passed review.
When a claim is superseded, the old claim is not deleted — it is marked as superseded with a link to the replacement. When a claim is retracted, downstream claims that depended on it are marked as stale. The history is preserved. The chain is traceable.
Collaboration
Knowledge is collaborative. Two engineers write the same design document. An analyst runs a notebook while a reviewer comments on the methodology. A design review happens in the 3D viewer with everyone marking up the same model. MUKR handles real-time collaboration natively — not as a bolted-on feature, but as the core data model.
CRDT-native means conflict-free. Two people type in the same paragraph — both edits survive. No locking. No 'document is being edited by someone else.' No last-save-wins. The Loro CRDT engine handles merges at the character level. LiveKit data channels deliver updates with p99 under 100 ms.
Presence is built in — cursors, selections, and avatars show who is where in the document. Screen sharing, voice, and video run through the same LiveKit session. A design review happens in the tool, not in a separate video call with a shared screen.
Universal Cross-References
Knowledge does not exist in isolation. A design rationale references a part in your PLM. A risk assessment references a transaction in your ERP. A compliance document references a regulation. MUKR's universal reference system links knowledge to the systems it describes — without stealing authority from those systems.
Type @ and reference any entity across your operations platform (parts, BOMs, work orders, quality records), your object platform (datasets, models, code), your ticketing system, your chat messages, and your document library. The reference is live — change the entity and the document reflects it.
Resolution follows a tiered path: local cache, knowledge graph, heuristic matching, object platform, operations platform, and external connectors. Unresolved references return AccessDenied — not NotFound — so that existence is not leaked to unauthorized users.
Who Uses MUKR
Engineer
Write design rationale with evidence, alternatives, and review state tied to parts and BOMs. When you leave, the rationale stays — with its evidence chain intact. Your replacement starts from the knowledge graph, not from a blank page.
Quality Manager
Link NCRs, CAPA records, and inspection results to claims with confidence scoring. When an auditor asks for the evidence behind a quality decision, the provenance export generates the package automatically.
Compliance Officer
Every claim, every decision, every policy interpretation carries its evidence chain. When a regulation changes, every claim that cited the old version is flagged for re-review. Audit-ready without manual reconstruction.
Researcher
Competing hypotheses coexist with their evidence. Notebook cells with reproducible provenance ensure analyses are traceable. When a peer reviewer asks how a result was produced, the notebook is the evidence.
Deployment Freedom
Knowledge is some of the most sensitive data in your organization — design rationale, compliance evidence, intellectual property, classified analysis. MUKR runs wherever you need it to run — cloud, on-premises, embedded in your operations platform, or fully air-gapped.
MUKR is available as a managed SaaS — we run the platform, your team writes the knowledge. Tenant-isolated with cryptographic separation. No cross-tenant leakage. When you want to bring it on-prem, the migration is a data export — not a re-platform.
Knowledge Portals
Internal knowledge is valuable. Customer-facing knowledge is a product. MUKR's 35+ component types deploy into the experience platform — so the same knowledge your teams write internally becomes branded, interactive, customer-facing portals without a separate content management system.
35+ component types: knowledge graph viewers, entity cards, wiki embeds, 3D model viewers, notebook cells, search bars, hover modals, chart renderers, verification badges, annotation overlays, and more. Each component reads from MUKR knowledge sources and binds to live data.
Components are permission-aware — the same component shows different content to an authenticated customer than to an anonymous visitor. Read-only and authoring modes depend on the user's role and realm.
Knowledge Surfaces by Usage Volume
Where knowledge is captured — documents dominate by volume, but 3D annotations and notebooks carry the deepest evidence. Graph explorers reveal relationships. AI composition synthesizes new knowledge from existing claims.
Every surface feeds the same claim graph. No knowledge is orphaned. No evidence is lost across tool boundaries.
Claim Lifecycle Distribution
Claims move through 11 states. Most claims reach Verified. Some stay Proposed. Disagreements are preserved, not lost — every dispute is a first-class object with evidence, context, and resolution state.
No claim is deleted. No evidence is discarded. The knowledge graph remembers everything.
Four Layers of Knowledge
Semantics, ontology, epistemology, and heuristics — each layer adds a different kind of intelligence. Semantics defines meaning. Ontology structures relationships. Epistemology tracks evidence and belief. Heuristics applies operational wisdom.
Together, they form a knowledge system that understands not just what things are, but why they are, how we know, and what to do about it.
Use-Case Surfaces
- Design Rationale CapturePreserve the why behind parts, assemblies, requirements, risks, and trade studies — with evidence, alternatives, and review state.
- Engineering Knowledge BaseWikis, documents, and 3D annotations connected to PLM parts, BOMs, and quality records — no more losing rationale across tool boundaries.
- Computational AnalysisReactive notebooks with Python, Rust, and SQL — analysis results become knowledge claims with reproducible provenance.
- Compliance EvidenceEvery claim, every decision, every policy interpretation carries its evidence chain — audit-ready without manual reconstruction.
- Customer Knowledge PortalsPublish curated internal knowledge into branded, interactive customer portals via the experience platform.
- Field Service GuidesInteractive call scripts and 3D part viewers for technicians in the field — with live data from your operations platform.
Why Your Knowledge Disappears
The engineer who designed the part leaves. The analyst who wrote the risk assessment moves teams. The compliance rationale behind a policy decision gets buried in email. The data remains — in your ERP, your PLM, your CRM, your ticketing system — but the why disappears. You can find the part number. You cannot find why the part exists, which alternatives were rejected, or which defect history changed the decision.
MUKR is built on a different premise. Knowledge is not a storage problem — it is an evidence problem. A claim is not knowledge because someone wrote it. A claim is knowledge because it has evidence, because it passed review, because its confidence is calibrated, because its provenance is traceable, and because its lifecycle state is governed. When an engineer leaves, the knowledge stays — with its evidence, its confidence, and its review chain intact.
That is what MUKR is. Not another wiki. Not another document store. An evidence system for institutional knowledge.
Proof points
- Every claim carries evidence, confidence, provenance, and lifecycle state.
- CRDT-native collaboration means knowledge capture happens in real time, without conflicts.
- Cross-references link knowledge to operational systems without stealing authority.
Objections answered
- MUKR does not own ERP, PLM, or CRM state — it provides knowledge context.
- MUKR is not a workflow engine — it tracks review state, not process execution.
- MUKR is not a search engine — it is an evidence system that happens to be searchable.
Operator relevance: A reviewer can answer whether a claim is usable, what evidence backs it, what has changed, and where escalation belongs — in one query.
Core Commitments
- RuntimeRust-native with Protobuf contracts and CRDT collaboration
- CollaborationLoro CRDT engine with Yjs bridge, LiveKit real-time transport
- Block Types35+ schema-validated block types for documents and portals
- Notebook KernelsPython, Rust, SQL, MQL, and Markdown with reactive dependency DAG
- 3D FormatsGLTF, OBJ, STL, STEP, IGES with WebGPU rendering at 60+ fps
- Reference SystemUniversal @mentions across operations, object, and external systems
- Epistemology7-dimensional confidence, Brier score calibration, 11-state claim lifecycle
- ProvenanceW3C PROV-compliant provenance graph with queryable, versioned, exportable chains
- Truth MaintenanceDependency propagation, stale detection, disagreement preservation, supersession chains
- SearchTantivy in-process full-text search across documents, wikis, and graph nodes
- Knowledge GraphAuto-extracted entities, inferred relationships, 2-hop queries under 50 ms on 100K entities
- DXP Components35+ component types for customer-facing knowledge portals
- DeploymentStandalone SaaS, embedded in operations, embedded in experience, hybrid, air-gapped
- OfflineLocal CRDT persistence with conflict-free merge on reconnect
- Performance10K-block document under 200 ms, LiveKit sync p99 under 100 ms, 3D 60+ fps
Platform
- RuntimeRust-native with Protobuf contracts
- CRDTLoro primary, Yjs bridge, LiveKit transport
- DeploymentSaaS, embedded, hybrid, air-gapped
Knowledge Surfaces
- Documents35+ block types, schema-validated, drag-and-drop
- Notebooks5 kernels with reactive dependency DAG
- 3D ViewerWebGPU, GLTF/OBJ/STL/STEP/IGES, 60+ fps
Epistemology
- Confidence7-dimensional state vector with Brier score calibration
- Lifecycle11 states from Proposed to Operationalized
- ProvenanceW3C PROV-compliant, queryable, exportable