Features

See how Lantide makes AI analysis reviewable, traceable, and reusable

Methodology

For analysts and PMs

Lightweight SQL Q&A vs structured project workflow. Simple questions don't force a mode; complex work escalates to projects.

Plan is pre-execution contract; Report is post-execution deliverable. Executed Plans are locked; new analyses get new Plans.

Annotations bind to source text; Resolve Comments drives Agent's next revision.

Generate/Re-generate full HTML, browser sharing, Quick Edit and Live Preview.

Plan execution offers Foreground and Hybrid modes; pipeline stages use SQL tabs to keep evidence visible and traceable.

Keep project background, definitions, and decision context in reference documents shared by people and Agents and carried with the project.

User/Project tiers, Rules/Info split, Queued Knowledge approved before injection.

For technical evaluators

State routing, state-filtered tool matrix, exploration budget soft cap.

Summary/Detail dual channel, layered prompts, annotation and knowledge injection optimization.

sqlglot rewriting, tab cache, Source Run DAG, deep lineage tracing.

Local CSV/Excel/Parquet, PostgreSQL/MySQL ATTACH, MCP Parquet materialization.

Read-only guard, human-only Execute, memory fully closable.