OpenDAA/OpenAnalytics

OpenAnalytics Semantic Decision Workbench

Build lean analytical pipelines that bind data, ontology, models, simulations, notebooks, evidence, and operational action. Semantics constrain execution before an LLM or robot gets the chance to improvise creatively.

OPEN OPENDATA OPEN GRAPH PIPELINE BUILDER
SEMANTICS → EVIDENCE → DECISION → TASK Safe built-in analytics run here. Open WebUI calls require an allowlisted connector. Jupyter, custom code, physics models, wargames, and privileged automation are claimed by an authenticated OpenAnalytics Agent. Operational effects flow through explicit authority and OpenTask.
PIPELINES—— active
RUNS—— queued
RELATIONSHIP GRAPHS—— nodes
MODELS—built-in + external
JUPYTER—— servers
DECISION RULES—advisory
EXECUTION ARCHITECTURE

Semantic execution loop

LIVE
→ → → → 5ACTCreate governed OpenTasks and automation intent
COMPUTE CONNECTIONS

Execution boundary

START WITH A CAPABILITY

Lean analytic building blocks

No 47-step wizard. Humanity survives.
OPEN OPERATIONAL ANALYTICS

From heterogeneous data to controlled action

OPENDATASQL • lake • search • graph • streams
→
OPENKNOWLEDGEontology • constraints • object identity
→
OPENANALYTICSpipelines • models • notebooks • simulations
→
OPENBUSreal-time events • replay • triggers
→
OPENTASKapproval • assignment • execution • audit
↔
OPENCOP / ROBOTSoperational picture • sensors • effects
ANALYTICAL DEFINITIONS

Pipelines

+ VISUAL PIPELINE
Select a relationship graph—
FULL OPENGRAPH
Build a graph to visualize entities, links, provenance, and data lineage.
JUPYTER SERVER INTEGRATION

Notebooks and kernels

EXECUTION TARGETS

Jupyter servers

ANALYTIC ASSETS

Registered notebooks

AGENT CONTRACTThe reference agent executes notebooks with jupyter nbconvert --execute or an approved runner, records the notebook path and environment, and reports results to the run API. Tokens remain in the compute environment, not in this registry.
MODEL & SIMULATION REGISTRY

Open WebUI, custom models, and simulations

INTERCHANGEABLE ADAPTERS

Registered models

SEMANTICALLY GROUNDED LLM

Open WebUI assistant

EVIDENCE REQUIRED
The LLM receives a bounded semantic context. Unsupported evidence IDs are flagged before the result can become an OpenTask.
REAL-TIME ANALYTICS & AUTOMATION

Semantic Decision Engine — Decision rules

ADVISORYCreate findings only
→
HUMAN APPROVALRecommend tasks for operator review
→
AUTO TASKCreate an OpenTask only when server policy and rule approval both permit it
→
ROBOT / SYSTEMExecution remains with the assigned agent and its safety controls
PROVENANCE & EVIDENCE

Analytic runs

RUNPIPELINEKINDSTATUSREQUESTEDACTION