Data & rules01
Data everywhere. Rules only in your head?
Connect CSV, Excel, Parquet, and external sources so agents can use the right data, definitions, and mappings.
The analytics workspace for the Agent era
Make AI analysisreviewable traceable reusable
Make data sources, analysis context, metric logic, and reports into analysis assets for reliable human–AI collaboration.
Free desktop app · macOS / Windows
01_plan.mdAnalysis plan excerpt
Order funnel analysis
The question
Where do orders drop off most in the last 90 days?
Denominator & grain
Orders created in the period. One row per order.
Approved → Carrier
status = 'shipped'Keep the revised calculation logic
carrier_events.handoff_at IS NOT NULLSourcesorders.csvcarrier_events
One plan, carried through review, execution, and reporting.
Explore the full workflowRerunnable computation02
One metric changed. Redo the whole analysis?
Preserve SQL, dependencies, and intermediate results so people and agents can see the impact and continue from the changed step.
Complete analysis assets03
The report is done. Explain it all again next time?
Keep plans, execution, evidence, and conclusions together so teammates or agents can verify, reuse, and continue.
01 / The method
From question to reviewable result
Start with quick exploration. Add structure when the analysis needs to be reviewed, rerun, or shared.
Start with quick exploration, then move into a focused project when you need a reviewable Plan → Report workflow.
1
Quick Analysis
Data exploration and problem framing
2
Project Analysis
Focus a project for deep-dive analysis
01
Plan
Write an analysis plan: what to query, which tables to use, and how to validate
02
Review
Review comments stay attached to the source; the Agent updates the analysis plan
03
Execute
After human approval, the Agent may perform analysis while keeping an ongoing analysis log
04
Report generation and continued analysis
Put conclusions and evidence into a document, export an HTML presentation, or continue into the next analysis cycle
02 / In practice
Use Cases
Structured Analysis Projects
A/B tests, funnels, anomaly detection — Plan/Report versioning
Privacy & Compliance
Data stays local, memory closable, no auto vector DB
Stakeholder Deliverables
Markdown Report + standalone HTML for bosses or clients
I already use Claude Code or Codex. Do I still need Lantide?
You do not have to replace your preferred Agent. MCP Server lets you keep working in Claude Code or Codex while Lantide retains the Plan, SQL evidence, Report, and Activity for review, metric changes, and handoffs.
Can I reproduce a number from six months ago?
Lantide retains the Executed Plan, SQL evidence, Report, Source Run lineage, and Activity so you can recover the original definition, understand dependencies, and rerun the analysis.
Where is my data stored?
Local-first design. Data is processed in your local DuckDB workspace and is not auto-uploaded to the cloud.
The full version is free — Start now
Download Lantide Data and experience reviewable data analysis workflows.