Sidekick Lab – AI-Powered Data Discovery Platform

Sidekick is an on-premises, inside-tenant AI platform that gives enterprises a living, governed understanding of their entire data estate, automatically. It connects to data source systems including Snowflake, Databricks, SAP, PostgreSQL, MySQL, Oracle, MongoDB and SQL Server, discovers every data source, field and table, builds a plain-English data dictionary and ontology, classifies sensitive data, and makes it queryable in natural language.

The core problem Sidekick solves: 73% of enterprise AI projects fail, most often because of poor data quality, unknown assets and a data estate that nobody fully understands before the project begins. Most organizations believe 80% of their data estate is documented, when the reality is closer to 30%. Sidekick closes that gap before any AI initiative begins.

• Sidekick is not a chatbot or a copilot that only answers the questions it is asked. It is an autonomous agent that discovers, audits, analyzes, maps and reasons over data without waiting for a user query, and improves its understanding of the estate continuously over time.
• Sidekick runs on fully supported Microsoft reference stack.
• Deploys in your tenant, burns your Azure commit/credits, solves your data-discovery and ontology problem.

What Sidekick delivers per source system: a data dictionary, sensitivity and PII classification, a cleanliness assessment, use case identification and ontology mapping. Across the organization it delivers an enterprise data catalogue, a cross-source interoperability map, an AI-readiness report, data monetisation opportunities and a proposed data strategy and roadmap.

Deployment and security:

Sidekick deploys fully inside the customer's own environment (Azure, on-premises, or other approved region), with no data leaving the customer's perimeter. Access to source systems is read-only with zero write access. The platform uses data minimisation, accessing only schemas and configurable sampled records rather than full extractions, and customers control scope using allow-lists, deny-lists or schema/table restrictions. Access control is managed through Microsoft Entra ID with role-based and field-level security. Full audit trails are available for SIEM integration, and the architecture is POPIA and GDPR aligned, HIPAA-suitable and built to ISO 27001 and SOC2 practices. Encryption is AES-256 at rest and TLS 1.3 in transit. No customer data is used for model training and no cross-tenant data sharing is possible.

Infrastructure:

Sidekick runs on a Windows VM (SQL Server 2022, minimum 4 vCPU / 16GB RAM) and a Linux VM (Docker containers via Docker Compose, minimum 2 vCPU / 8GB RAM). It is strategically aligned with Microsoft, running on Microsoft AI Foundry and Azure, and extending into Microsoft Fabric and Copilot. The platform is cloud-agnostic and can also run on AWS or GCP, and a partner-hosted LLM option is available for fully air-gapped environments.

Current agents include Discovery and Cataloguing (scans the full environment, builds and maintains the catalogue, identifies cross-source use cases) and Reporting and Insights (natural language questions with real-time answers, shareable dashboards, improves with feedback over time).