Invofox

Purpose

Invofox's Custom Docs feature is aimed at teams that need to instantly extract and validate data from any document. As long as a document contains describable fields, it can be structured, for example insurance assessment reports, car rental handover forms, handwritten harvest logs, subcontractor timesheets, and cold storage temperature records. The README emphasizes: "If your documents have fields, we can extract them."

Main capabilities

• Supports user-defined schemas, extracting any fields, and validating before the data flows to downstream systems.
• Uses an agentic pipeline to automatically adapt to documents, learn from a small number of samples, and apply the user's validation rules.
• Delivers structured JSON via an asynchronous API and webhook; the output can be directly connected to ERP, CRM, BI, or custom applications without additional parsing.
• Supports treating layout variants from different branches or end customers as part of the same schema, without setting up each branch separately; the pipeline automatically routes and adapts.
• Schemas support versioning and editing, and newly added fields later do not require re-onboarding.
• The official statement says standard document types and special documents use the same endpoint, with the same quality and the same price.

How to use

According to the official FAQ:

1. No model training is required on the user side. Invofox will ask for a set of representative document samples to configure and fine-tune its Agentic Pipeline for the user's schema. The official explanation says the base model already covers most layouts, and the samples are used to tune routing, validation, and field-level behavior.
2. The number of samples required ranges from one to several dozen, depending on layout variability, and the official team will tell you upfront after seeing the first sample.
3. Once the schema is defined and the documents are available, most customers run their first production batch in about a week; after that, accuracy continues to improve in subsequent batches through an automatic feedback loop.
4. Data is delivered as structured JSON via an asynchronous API and webhook.
5. The official team offers a "send a sample, show the output" approach, so you can try it without a schema, integration, or commitment.

Applicable scenarios

• Documents that do not fit any template, such as internal forms, supplier paperwork, and site logs.
• Insurance assessment reports: on-site assessor notes, loss estimates, and signed valuations.
• Car rental handover forms: vehicle condition snapshots taken by counter staff.
• Handwritten harvest logs: daily handwritten field records from orchards.
• Subcontractor timesheets: hours and tasks recorded by foremen, varying by contractor.
• Cold storage temperature records: hourly readings from warehouses across the cold chain.

Limitations

The materials do not explain specific pricing, field quantity limits, supported languages, data retention policies, or compliance certifications. The document mentions "the same price," but does not provide specific pricing information, so it is not available for now. The questions and answers in the official FAQ are presented as example JSON, and some details, such as the specific technical process of tuning, are not expanded on.