Skip to content

Data sources — what we connect to

Last updated: 2026-05-14

Customer data lives in many places. The Base2ML platform is designed to accept content from wherever it actually lives, with the integration scoped to the customer's specific environment rather than a generic "all storage systems" promise.

Available out of the box

Source How it works Setup time
Drag-and-drop upload Web UI, multi-file, supports PDF, DOCX, TXT, MD, CSV, XLSX, EML, MBOX, PNG, JPG, TIFF 5 minutes
CSV bulk import Any tabular export (job records, equipment logs, customer history, ticket history). Each row indexed as a queryable record 10 minutes
Email archive ingestion MBOX or EML files (Gmail Takeout, Outlook export, IMAP backup) 10 minutes

Connector on request — scoped to your environment

Built per customer, typically within 1 week of discovery. We scope to your specific tenant configuration rather than maintaining a generic connector.

Source Notes
Google Drive OAuth + service account, scoped to specific folder(s)
Microsoft SharePoint OAuth via Microsoft Graph API, scoped to specific site / library
Microsoft OneDrive Same auth path as SharePoint
Box OAuth, scoped to specific folder hierarchy
Dropbox / Dropbox Business OAuth, scoped to team folders
Network file share (SMB / NFS) Requires VPN access or sync script
AWS S3 / Google Cloud Storage / Azure Blob Cross-account access patterns; same-region or cross-region
GitHub / GitLab / Bitbucket OAuth, scoped to specific repos; ingests README, docs, wiki content

Job-management & CMMS systems — scoped integration

We don't maintain a connector library for every CMMS / job-management system. We build the connector that fits your specific system once we know what you use.

Category Examples Approach
CMMS (maintenance) Maintenance Connection, eMaint, Limble, Fiix, UpKeep, Hippo, MaintainX, MicroMain API + scheduled sync OR CSV export pipeline
EAM (enterprise asset mgmt) SAP PM, IBM Maximo, Infor EAM, Oracle eAM Typically CSV export pipeline; API requires customer IT engagement
Service / job management ServiceTitan, FieldEdge, Sera, Jobber, Housecall Pro API-based scheduled sync
Project management Procore, Buildertrend, CoConstruct, Builder Trend API-based scheduled sync where API exists
ERP / accounting NetSuite, QuickBooks, Sage, Microsoft Dynamics Typically read-only API for specific record types

Honest disclosure: "scoped to your environment" means we build it for you, not that we have it pre-built. The first 1-2 customers on a system effectively contribute to the connector we maintain afterward. For your engagement we'll quote the integration time on the discovery call.

Quick-start option — the watched-folder sync script

For customers who want to evaluate quickly without a full integration build, we provide a small sync script that runs on one of your team's machines, watches a folder you pick, and auto-uploads new files to the Base2ML platform.

Works with anything that appears as a folder on a desktop:

  • Google Drive Desktop (sync'd folders)
  • OneDrive Desktop
  • SharePoint synced folders
  • Box Drive / Dropbox Desktop
  • Any network drive mounted on the machine
  • Any local folder

Setup: Download script, point at folder, configure credentials. ~15 minutes. Limitations: Single user / machine; not real-time (configurable polling interval, typically 5-15 minutes); requires the machine to be running.

Good fit for: pilot engagements, smaller corpora, teams where one person curates the searchable content.

Not a fit for: real-time critical workflows, large-scale or multi-user-editor environments — use a direct connector instead.

How we decide what to build

On the discovery call, we ask:

  1. Where does the content actually live today? Specific systems, specific tenants.
  2. Who curates it? Is there a single source of truth or multiple?
  3. How often does it change? Daily edits or static archive?
  4. Who needs to search it? 5 users or 50?

Those four answers determine whether you need a one-off CSV import, a watched-folder sync script, or a full OAuth connector — and we scope the time accordingly. The discovery answer drives the build, not the other way around.

What we don't do

  • We don't sell "connectors to every storage system on the market." That's a positioning lie that catches up to you on the first edge case.
  • We don't build connectors speculatively. Every connector we maintain has at least one paying customer behind it.
  • We don't pull data that requires elevated privileges beyond what you grant. Read-only OAuth scoped to a specific folder / library / repo is the default; never broader.

Contact

Email chris@base2ml.com with what you use today. We'll scope the integration on the discovery call.