Telmai Data Observability

Telmai data testing

Telmai checks SQL tables, and its Data Diff finds new, missing and changed records between two datasets, storing per-record differences.

Compare: data comparison tools

• SQL Table Tests: Record Validation Rules check individual records against completeness, validity, and uniqueness business rules; SQL-Based Rules support arbitrary pushdown SQL expressions including multi-table joins for consistency checks (docs).

• Data Diff: The Data Diff feature compares two datasets to identify new, missing, and changed records, storing per-record diffs as Parquet files and firing alerts on detected differences (docs).

• DataFrame Tests: Telmai connects to data sources via SQL connectors and cloud storage; no Python DataFrame (Pandas, Spark, Polars) in-memory validation library is provided (docs).

• Unstructured Data Tests: No LLM-based semantic validation of free-text or JSON warehouse columns found in documentation; validation rules are expression- or SQL-based (docs).

• Data Contracts: No data contract feature (machine-readable producer-consumer schema and quality SLA agreements) found in documentation (docs).

• Table Comparison Tests: The Data Difference out-of-box monitor compares data between sources/periods but appears to operate at the row level; no explicit aggregate comparison test (row count match, sum match between two tables) was confirmed (docs).

• Inspect Failed Rows: Individual failing rows are not explicitly surfaced in the Incident Portal UI; Data Binning separates valid/invalid records into separate bins but row-level inspection within incident triage is not confirmed (docs).

Telmai data observability

Telmai's monitors learn normal patterns over the first monitoring cycles and adapt their thresholds. Monitors are configured in the UI, and its Investigator lets you explore distributions and patterns.

Compare: data observability tools · data profiling tools

• Data Profiling: The Profiling Data section includes Data Health Metrics, an Interactive Profiling Tool (Investigator), and Compound Attributes for column-level statistical analysis (docs).

• Anomaly Detection: Telmai establishes normal patterns over initial monitoring cycles and continuously compares current metrics against learned baselines, adapting thresholds over time as data evolves (docs).

• Freshness Monitors: Freshness Drifts out-of-box monitor tracks the time elapsed since the last table update and alerts when the interval exceeds the learned baseline (docs).

• Volume Monitors: Record Count Drifts out-of-box monitor detects unexpected changes in data volume using ML-derived baselines (docs).

• Schema Monitors: Schema Drifts out-of-box monitor detects added or removed columns and fires alerts on structural changes (docs).

• Field Monitors: Completeness Drifts, Uniqueness Drifts, and Numeric Value Drift monitors track column-level statistics as continuous time series with ML-based anomaly detection (docs).

• Distribution Monitors: Data Drifts, Value Distribution Drifts, and Top Values Drifts out-of-box monitors track statistical distributions and category frequencies per column over time (docs).

• Custom SQL Monitors: SQL-Based Rules support arbitrary pushdown SQL queries — including multi-table joins — tracked as continuous monitors with configurable thresholds (docs).

• UI-Based Monitors: All monitor types can be created and configured through the Telmai UI via the monitor creation panel; the tool is marketed as a no-code/no-SQL platform (docs).

• Monitors-as-Code: No YAML or code-first monitor definition approach found; monitors are managed through the UI or programmatically via the REST API, not as version-controlled code artifacts (docs).

• Pipeline Monitoring: No pipeline job runtime monitoring features found in documentation (docs).

• Segment Monitors: Data Binning separates records into valid/invalid bins, but dimension-based segment monitoring (break down metrics by region, product_type, etc.) is not explicitly confirmed (docs).

• Partition-Based Monitors: Partition-specific scanning (e.g., scan only yesterday's date partition) is not explicitly documented (docs).

• Window-Based Monitors: Monitors reference past scan history via 'Compare to Past # Scans' but configurable tumbling or sliding time-window evaluation is not explicitly documented (docs).

Telmai incident management

Telmai sends alerts, tracks incidents through triage and resolution, and opens tickets.

• Alert Notifications: Slack is documented as a notification channel for alert delivery; monitor creation includes a Notification section for configuring endpoints (docs).

• Incident Triage: The Incident Portal provides a dashboard with incident cards (open/closed counts, MTTR), severity-based impact visualization, and a sortable incident table for structured triage (docs).

• Incident Resolution: The Incident Portal tracks incidents through open and closed states and shows MTTR; incidents can be closed and their full alert history is accessible (docs).

• Ticketing Integrations: Native Jira integration automatically creates Jira issues from Telmai incidents and alerts (docs).

• On-Call Integrations: No PagerDuty or Opsgenie on-call paging integrations found; documented notification integrations are Slack and Jira only (docs).

• Alert Feedback: No explicit false-positive marking, alert muting, or ML feedback loop on individual anomaly signals found in documentation (docs).

Telmai data quality analytics

Telmai's dashboards, health scores and resolution metrics track data quality.

• Data Quality Dashboards: DQ Report Summary, Data Trends and Alerts, and the Incident Portal provide built-in dashboards for quality KPIs, trend charts, and incident tracking (docs).

• Data Health Scores: The Data Quality Score computes a per-dataset health score (0–100) as a weighted average across Completeness, Validity, Freshness, and Integrity dimensions with configurable weights (docs).

• Incident Resolution Metrics: Telmai's Incident Portal dashboard shows open/closed incident counts, an incident-volume-over-time trend graph, and mean time to resolve (MTTR) as summary cards (docs).

Telmai data governance

Telmai classifies data and syncs results to Alation, Unity Catalog and Hive Metastore.

• Data Catalog Integration: Alation integration pushes Telmai data quality results into the Alation data catalog (docs).

• Data Classification: PII Exposure out-of-box monitor detects and classifies personally identifiable information in data attributes, generates alerts and incidents, and supports custom sensitive data validation rules with data binning remediation (docs).

• Data Catalog: Telmai has no built-in searchable data catalog; the Assets section manages monitored datasets but does not provide catalog-style discovery with descriptions, tags, or business glossary (docs).

• Data Products: Telmai groups data sources under "Projects," which are workspace/tenant containers for managing access and organizing sources, not a data-mesh-style domain or data-product grouping of monitors/quality scores (docs).

Telmai workflows

Telmai has a REST API, webhooks, scheduling and an MCP server. It has no Python SDK or dbt integration.

• Scheduling: Scans can be scheduled to run automatically at a specified time via the Telmai UI or triggered programmatically via the Upload Data API (docs).

• Webhooks: Jobs Status Notification delivers webhook callbacks to a user-configured endpoint when dataset processing jobs change state (docs).

• REST / GraphQL API: Full REST API covering connections, assets, monitors, incidents, scan triggers, DQ reporting, and ticket integration operations (docs).

• MCP Server: The Telmai MCP Server connects AI assistants (Claude, VS Code Cline) to Telmai data quality context via the Model Context Protocol (docs).

• Python SDK: No pip-installable Python SDK found; programmatic access is available only via the REST API (docs).

• dbt Integration: No dbt integration (manifest import, test syncing, or dbt Cloud connection) found in Telmai documentation (docs).

• CI/CD Integrations: Circuit Breaker feature switches off ETL operations via API when bad data is detected, but a first-class CI/CD quality gate that automatically blocks pipeline deployments is not documented (docs).

Telmai AI agents

Telmai writes checks from plain-language prompts and suggests checks automatically. An MCP server is also available.

• Natural Language Test Generation: Telmai's homepage states its Data Reliability Agents let "both technical and business users set up data monitors, query data quality issues and validations in plain English," i.e. deploying monitors from plain-language commands (website).

• Automated Test Generation: Out-of-box monitors (freshness, volume, schema, field-level, distribution) are automatically enabled on connected datasets without requiring user configuration (docs).

• AI Chat Assistant: Telmai has no in-product AI chat assistant; NL capabilities are limited to the external MCP server (for AI agents) and the Validation Agent which auto-suggests rules from data trends without user interaction (docs).

• AI Issue Triaging: The Incident Portal tracks incidents and displays history with no ML or LLM reasoning. No AI-powered root cause analysis or issue classification feature is documented (docs).

• AI Issue Resolution: Remediation consists of Data Binning and Circuit Breaker — both are manual decision frameworks with no AI-proposed or automated fix execution. Circuit Breaker requires the user to decide the best action after checking health KPIs (docs).