Telmai Data Reliability Agents

Telmai is an AI-based data observability company registered in the United States with an address in San Francisco. It focuses on data that lands in data lakes and open lakehouses, where structured, semi-structured and unstructured data must be validated before it feeds analytics and agentic AI workflows.

The platform runs on a Spark engine at petabyte scale and validates data at the lake layer, including Apache Iceberg and Delta Lake tables, detecting anomalies, drift, schema, freshness and completeness issues as data arrives. Seven Data Reliability Agents handle orchestration, rule suggestions, incident diagnosis, lineage mapping and insights through natural language, and quality metadata is exposed to AI agents via MCP. A certified Microsoft Fabric workload runs computation inside the customer's Fabric tenant. PropertyGuru Group is among the named customers.

• Lake-layer validation: Checks structured, semi-structured and unstructured data as it lands, including Iceberg and Delta tables.
• ML anomaly detection: Monitors volume, schema, freshness, completeness and drift without manual thresholds.
• Data Reliability Agents: Seven AI agents for orchestration, rule suggestions, incident diagnosis, lineage and insights.
• Natural language interface: Set up monitors and query quality issues in plain English.
• Incident management: Correlates issues, explains root causes and suggests resolution steps.
• MCP trust signals: Machine-readable quality metadata that AI agents and catalogs can consume.

A Vietnamese e-commerce or property platform can validate clickstream and listing data landing in an Iceberg or Delta lakehouse before it feeds recommendation models. A bank using Microsoft Fabric can monitor OneLake data products for freshness and completeness and block AI agents from using stale data.

Quality checks run at the lake layer before bad data reaches dashboards or AI agents, covering open formats like Iceberg and Delta Lake. Business users can set monitors and investigate incidents in plain English, reducing dependence on data engineers. In Microsoft Fabric, raw data stays in the customer's tenant and only derived metrics are stored. A 30-day trial and usage-based pricing via Azure Marketplace lower the entry barrier.

Enterprises are moving analytics to open lakehouse formats and connecting data to AI agents, which increases the risk of unvalidated data reaching automated decisions. Southeast Asian digital platforms, such as PropertyGuru, are adopting lakehouse observability, and many enterprises in the region are evaluating Microsoft Fabric where Telmai runs natively.

Telmai lists its address at 2261 Market Street, San Francisco, CA. Its Telmai Data Reliability workload passed Microsoft Fabric Workload Hub publishing requirements, and the company reports a SOC 2 Type 2 audit. It holds GigaOm leader badges and G2 2025 badges. Partners include Microsoft and iLink Digital.