Built for AI agents
Chartstone is an HTTP API, but the experience it unlocks shows up
most clearly when you point an AI agent — Claude Code, Cursor,
ChatGPT’s desktop tools, your own custom agent — at it.
Ask a question in plain English, the agent translates it into the
right call, Chartstone runs it against your live NetSuite session.
Ask
“Show me the top 10 customers by total invoice amount this
year.”
Chartstone delivers
The agent writes SuiteQL aggregating invoices by customer, posts
it to /suiteql, and gets ranked rows back in
seconds — no saved search, no report design, no preconfigured
endpoint required.
Ask
“Who changed any credit limit in the last 30 days, and
what to?”
Chartstone delivers
The agent queries systemNote filtered to the
creditlimit field over the date range, returns a
full audit trail with old value, new value, user, and
timestamp.
Ask
“What fields does the salesorder record have? Which ones
are required?”
Chartstone delivers
The agent calls /records-catalog/schema for
salesorder, returns the full field list with types and required
flags. It can keep going — “and what about the line
items?” — without you naming any tables in advance.
Ask
“Run the A/R Aging Summary report for last quarter and
flag anyone over 90 days.”
Chartstone delivers
The agent uses /reports to find the report ID,
/report-info to inspect filters,
/report to run it with the right period, and
filters the result rows to overdue customers — all in one
conversation.
The architecture is what makes this fast and cheap: TOON output
keeps LLM context tokens small, schema is discoverable so agents
don’t need pre-registered tools per record type, and
everything runs on loopback so there are no API keys for the
agent to fumble.
See the full agent cookbook →