Train / Data connectors

Turn your systems into training data

Connect a source once and the training file keeps itself current. Databases, SaaS, APIs, files and documents become validated JSONL datasets a fine-tuning job reads by ID.

Connect your sources95 sources
SaaS
Salesforce · HubSpot · Zendesk
Databases
PostgreSQL · Snowflake · Databricks
APIs
REST · GraphQL · Webhooks
Files
CSV · Amazon S3 · Excel
Documents
Google Drive · Microsoft 365 · Confluence
Messaging
Slack · Gmail · Microsoft Teams
Analytics
Google Analytics · Google Ads · Datadog
Industrial
MQTT · OPC UA · Modbus
Generated datasetsRefreshed daily · 04:00
current
support-replies.jsonlPrompt → response
8,412
claims-triage.jsonlMulti-turn chat
5,067
onboarding-chats.jsonlChat → reply
1,908
returns-dialogue.jsonlQuestion → answer
2,640

The systems you already run

95 sources in eight kinds, from the catalogue: databases, SaaS, APIs, files, documents, messaging, analytics and industrial systems.

One file per source, kept current

Each source writes a JSONL file and refreshes it on the schedule you set, so the training set never goes stale.

Mapped onto the chat format

A template over the record's fields sets the system, user and assistant turns, with a filter for the records that qualify.

Validated before you spend GPU time

Every line is checked, and a rejected line comes back with its number and the reason.

How it works

Connect a source once. The training file keeps itself current.

Catalogue

Every source, filterable by kind

95 sources in eight kinds. Filter by kind or by what the system holds, search by name, and connect a source from its card. Each one writes a training file of its own.

  • Salesforce

    SaaS

    Cases, contacts and their resolutions

  • HubSpot

    SaaS

    Deals, contacts and the notes on each

  • Zendesk

    SaaS

    Support tickets and the replies that closed them

  • Jira

    SaaS

    Issues, comments and how each was resolved

  • SAP

    SaaS

    Master data, orders and their documents

  • Shopify

    SaaS

    Orders, products and customer messages

  • PostgreSQL

    Databases

    Tables and rows as records

  • MySQL

    Databases

    Rows from any query you write

  • Snowflake

    Databases

    Warehouse tables, filtered by a query

  • Databricks

    Databases

    Lakehouse tables and the results of a query

  • REST

    APIs

    Any JSON endpoint, paged and mapped

  • GraphQL

    APIs

    Records from a query against your schema

  • Webhooks

    APIs

    Events pushed to you as they happen

  • CSV

    Files

    Rows from a delimited file, with headers

  • Amazon S3

    Files

    Objects from a bucket, by prefix

  • Excel

    Files

    Sheets and named ranges as rows

  • Google Drive

    Documents

    Docs, sheets and slides, with their text

  • Microsoft 365

    Documents

    Documents, spreadsheets and site pages

  • Confluence

    Documents

    Wiki pages and the spaces holding them

  • Slack

    Messaging

    Channel conversations and threads

  • Gmail

    Messaging

    Threads, replies and their attachments

  • Google Analytics

    Analytics

    Sessions, events and traffic sources

  • Google Ads

    Analytics

    Campaign spend, clicks and conversions

  • MQTT

    Industrial

    Device telemetry by topic

At a glance

What the connectors give you

Sources

  • 95 sources in eight kinds
  • SaaS, databases and warehouses, APIs, files and storage
  • Documents, messaging, analytics and monitoring, industrial and IoT
  • Filter by kind and by what a system holds

Mapping

  • A template over the record's fields
  • System, user and assistant turns
  • A filter for the records that qualify
  • Purposes: fine-tune, evaluation and retrieval

Refresh

  • Hourly, daily, weekly or manual
  • Every refresh writes a new validated file
  • The file ID always names the latest ready version
  • History kept newest first, with rejected lines and reasons

Control

  • Credentials stored encrypted and never returned
  • Keys bound to another project cannot read a source
  • Disconnecting keeps the files until you delete them in Files
  • An Idempotency-Key guards a retried create
Pricing

Pay for what you keep

Storage

$0.10per GB / month

Datasets, checkpoints and exported weights are metered by the GB-hour at the monthly rate, so you only pay while you keep them.

Training tokens are metered separately when a job runs.

Pick your path

More in Train

Fine-tuning

Managed training on 25 open models

Explore Fine-tuning

Training loop

Docs

RL, DPO and distillation as API calls

Explore Training loop

Models

Open bases across six modalities

Explore Models

Connect your first source

Pick a source from the catalogue, map it, and set the refresh. The first file is ready in minutes.