DataLane

Open-source AI data analysis canvas. Connect CSV, JSON, or a live URL, describe a chart in plain English, and get an editable artifact — not a code dump.

DataLane is the open-source, self-hosted answer to AI data-analysis tools that only hand you generated scripts. It runs entirely on your machine, works with your own model key, and every chart is a real object you can edit, branch, and re-render.

What it solves

• You have data scattered across files and URLs and want charts without writing SQL or Python by hand.
• AI analysis tools give you a script you must re-run; DataLane hands you an editable HTML artifact with the spec + data embedded.
• Follow-up questions drift back to raw data; DataLane anchors a cleaned state so each question builds on the last.

Install

npm install -g @talocode/datalane
# or
pip install talocode-datalane

Quickstart

datalane analyze sales.csv "total revenue by month" --out revenue.html

from talocode import DataLaneClient

client = DataLaneClient()
source = client.load("sales.csv")
result = client.analyze(source, "average revenue by region")
print(result.summary)
client.to_html(result)  # editable artifact

CLI

datalane analyze <source> "<intent>" [--out <file.html>]
datalane render <source> --spec '<json>' [--out <file.html>]
datalane anchor <source> [--name <name>]
datalane from-anchor "<intent>" [--out <file.html>]
datalane branch --spec '<json>' --vars '<json>' [--out-dir <dir>]
datalane columns <source>
datalane sources | pricing | capabilities

API surface

Chart types

bar, line, scatter, pie, table

Aggregations

sum, avg, count, min, max

Sources

Intent parsing

Plain-English phrases are parsed deterministically — no key required:

• "total revenue by month" → bar, sum of revenue grouped by month
• "average spend by country limit 5" → avg, filtered to 5 rows
• "how many orders by region" → count
• "revenue trend over time" → line chart
• "breakdown by region as a pie" → pie
• "correlation between x and y" → scatter
• "where region contains South" → filter

Anchored state

anchor saves the cleaned dataset to .datalane/anchor.json. from-anchor and branch build on that clean state so follow-ups do not drift back to raw data.

Credits (hosted API)

Hosted at https://api.talocode.site/v1/datalane/*. Set TALOCODE_API_KEY and optionally TALOCODE_BASE_URL.

Related packages

• npm i @talocode/llmgateway / pip install talocode-llmgateway — LLM API gateway
• npm i @talocode/xsearchlane / pip install talocode-xsearchlane — X search
• npm i @talocode/doculane / pip install talocode-doculane — office documents

Talocode ecosystem

More: github.com/talocode · talocode.site · docs.talocode.site

License

MIT © Talocode