Dataiku AI Agents

Activities: Y/M	Diagnostics: P/M	Provisioning: P/L
Event ontology: Y/M	Observability: Y/M	Confidence: High

Build style / interface — No/low-code + Python; LLM Mesh; Trace Explorer.
What it actually does — Build, trace and evaluate governed agents; orchestrate data workflows.
Data / telemetry — Structured agent traces & metrics.
Interoperability — LLM Mesh; plugins; LangChain-style patterns.
Deployment model — Dataiku DSS.
Notes — Strong guardrails/tracing; infra provisioning out of scope.

UKM Snapshots: ingest Y/M, index Y/M, retrieval Y/M, governance Y/H, overall medium-high
Note: Ingestion & Validation: Y/M; Normalization & Enrichment: Y/H; Dataiku provides a unified environment for data, models, and now agents

UAM Snapshots:
ingest Y/M, index Y/M, retrieval Y/M, governance Y/M, overall medium-high
Note: Capture & Sessionization: Y/M; Evidence Policy & Verification: Y/H

UDM Snapshots:
ingest P/M, index P/M, retrieval P/M, governance P/M, overall medium
Note: Anomaly Detection & Monitoring: Y/M; Step Analysis: Y/H

UOM Snapshots:
ingest Y/M, index Y/M, retrieval Y/M, governance Y/M, overall medium-high
Note: Instrumentation & Ingest: P/L; Dataiku can ingest many data sources (databases, files, etc; Normalization & Enrichment: P/M; Dataiku strongly encourages schema and data quality checks

UEOM Snapshots:
ingest Y/M, index Y/M, retrieval Y/M, governance Y/M, overall medium-high
Note: Coverage — Ingest Y/M; Index Y/M; Retrieval Y/M; Governance Y/M; Overall medium-high

UPM Snapshots:
ingest P/L, index P/L, retrieval P/L, governance P/L, overall low-medium
Note: Coverage — Ingest P/L; Index P/L; Retrieval P/L; Governance P/L; Overall low-medium

Latest updates — AI Agents available across Dataiku with LLM Mesh (2025). Links — Official Site, Docs, Blog