Workspace
Business context, data and readiness in one place.
Create a decision workspace
Load a domain ontology, connect the data it describes, and the semantic runtime will determine what can be analysed.
Existing workspaces
Open an existing domain workspace.
Workspace
Business model
What this domain means.
View entities and compiler details
Connect data
Choose the governed data that should answer questions for this workspace.
Understanding
What the governed model can reliably interpret and calculate from the connected data.
Ready to analyse
Needs attention
Ready
Connect data to start asking business questions.
Explore your business
Metrics, dimensions and verified analyses from the active workspace.
Business metrics
Measures supported by the active data.
Dimensions
Ways the business can be broken down.
Suggested analyses
Business questions this ontology and data can support.
Technical & semantic tools
Detailed controls for semantic engineers, data architects and administrators. Everyday users should rarely need these screens.
Compile domain ontology
Only MECBot DDM export JSON (response.fc_entity / response.fc_attribute) is accepted. OWL, RDF, JSON-LD and generic graphs are rejected.
Compiled contexts
Immutable, versioned compilations. Governance is applied on load and survives recompilation.
Context summary
Entity graph
Directed MECBot references (source → target, cardinality, join key).
Compiler warnings
Ontology defects are reported, never silently repaired.
Semantic paths
Explore the fabric
Search any entity, field, metric, definition, OKF concept, document, view, use case or analysis, then walk its relations.
Relations
Recent analyses
For downstream apps
Other MECBot apps read the same governed understanding through the Metaspace API.
Cross-entity metric builder
Each component is aggregated independently at its native grain, rolled forward to a proven common grain, then combined. Approval is only possible when common-grain validation succeeds.
Metric horizon
Metrics reachable from an entity, grouped by native grain, with grain-safety status.
Ontology-bounded analytical closure
Finite set of valid (metric × operator × dimension × time) combinations — computed lazily, never materialised as millions of rows.
Intent grammar
Governed data runtime
Upload view profile
JSON produced by the MECBot Governed Runtime (view_id, doc_count, fields{type, exists_count}), an enterprise data index mapping, or a list of views. Profiles without statistics are marked SCHEMA_ONLY.
Data availability
Metrics and KPIs filtered by what the actual data supports. A metric is available only when every field it needs (measure, counted id, condition, components) has data in a profiled view.
Profiled views
Add enterprise knowledge
PDF, DOCX, PPTX, TXT, Markdown, CSV or JSON. Chunks are scoped to the active context with tenant, ontology version, section/page and linked entity/metric IDs.
Hybrid search
Exact semantics → governed vocabulary → graph expansion → BM25 → vectors. Vectors never override exact bindings.
Google OKF bundles
Open Knowledge Format v0.2 — Markdown concepts with YAML frontmatter. Upload a .zip, .tar.gz or .md files. Concepts are indexed for search; matches with the Semantic Context become pending proposals — nothing is applied until approved.
OKF proposals
Link external metric definitions (synonym + description), sanctioned computations, field descriptions and schema-only views. Deprecated or stale concepts are flagged.
What would you like to understand?
Ask in business language. FORMCEPT uses only the active workspace, governed semantics and connected data.
Saved analyses Reusable questions and approved analysis patterns
Business use cases derived from the ontology
Every use case the ontology can actually answer: goal × metric × dimension × time, plus framings proposed by Claude / OpenAI / Ollama — all planned by the deterministic engine, verified against governance and data, stored as OKF type: Use Case and vectorized into the semantic collections. A question is then answered by searching the collections, not by planning from scratch.
Semantic search over the collections
What the Metaspace matches a question against: use cases, intents, metrics, fields and OKF concepts.
Decision Board
Explore, map and curate business intents. Turn approved semantics into a governed Decision App.
Curate a board
A board assembles mapped intents, evidence and actions into a decision narrative. Every analytical output is re-evaluated against governed data when opened.
Decisions this workspace can support
Derived from the business model and the data actually connected. Add one and it starts watching.
Saved boards
Boards you composed from an answer or from the use-case catalogue.
Connect reasoning and build providers
Claude and ChatGPT subscriptions are used for Claude Code / Codex app generation. Decision Exploration remains governed by the semantic runtime, local Ollama, or an administrator-configured Claude/OpenAI API or enterprise gateway.
Governed deterministic and Ollama require no external subscription. Claude/OpenAI reasoning is configured under using enterprise/API credentials. Personal coding subscriptions are not used as a hidden reasoning API.
Turn a business question into a governed Decision App
One Decision Thread preserves verified understanding across Governed deterministic, ChatGPT/OpenAI, Claude and Ollama. The generated app inherits governed metrics, views, decision logic and lineage.
Decision Threads
Durable business context that can be explored, promoted to a Decision Board, and compiled into an application.
Published & Deployed Apps
Certified applications appear here throughout their lifecycle. Approve & Deploy publishes the immutable certified release when needed, verifies the governed runtime, and creates the running application without requiring a host-path or MECBOT_HOME choice.
Planner, embeddings and execution
Saved here, stored server-side, and used instead of the environment variables of the same name — no restart needed. API keys are write-only: they are never sent back to the browser. Leave a field empty to fall back to the environment (or the built-in default).
Review center
Everything this workspace is waiting on a person to decide — most consequential first, each one settled right here. Deciding an item says what it unlocked.
Advanced governance controls
Mapping approval policy
Controls whether term → semantic-object mappings produced by Ask Intelligence need a person to approve them before execution.
Business definitions
Rules derived from ontology evidence. They become authoritative only after approval.
Business vocabulary
What a word means when somebody uses it in a question. “Sales”, “orders”, “customers” each have to point at exactly one measure — otherwise a question that uses the word can only be answered by naming the measure instead. Deciding one here applies to every question, board and decision app from then on.
Approved / rejected mappings
Grain allocation rules
Governed resolutions for fan-out paths (distinct_parent / equal_split).