What do you want to do?#
Quick start (5 minutes)
Upload your first file and find it with AI search.
Uploading documents
Browser upload, capture email, API and link collections.
Document types & JSON schemas
Define what to extract from each kind of document.
Search
Keyword, semantic and natural-language search with the query builder.
AI features
Summaries, comparison, AI jobs and AI configuration.
API reference
Authenticate and push documents from your own systems.
How docuflow-ai works#
Every file goes through the same automated pipeline. You upload it (or email it, or push it through the API); background tasks split bundled PDFs, detect new versions, read the text with OCR, extract the fields your document type defines into JSON, and index everything in OpenSearch. When the pipeline is done the document is Searchable: you can find it by keyword, by meaning, or by asking a question in plain language.
IngestBrowser, capture email, API, link collections
PrepareDuplicate check, AI split, versioning
ReadOCR and text extraction for 75+ formats
ExtractAI maps content onto your JSON schema
UseSearch, compare, summarize, report
Read Core concepts & pipeline for the full picture, including every document status.
Key capabilities#
| Capability | What you get | Guide |
|---|---|---|
| Multi-channel ingestion | Drag-and-drop upload, a per-organization capture email address, a REST API and scheduled link collections. | Uploading |
| Broad format support | PDF, Office, OpenDocument, images, e-mail, e-books, markup and data files, with in-app preview. | File types |
| AI document splitting | Bundled PDFs (for example an order and its invoice) are split into separate business documents. | Concepts |
| Automatic versioning & comparison | Re-uploaded files become new versions; AI can compare versions and email the differences. | Documents |
| Schema-driven extraction | Each document type carries a JSON schema; AI fills it from the document's content. | Schemas |
| Keyword, semantic & AI search | OpenSearch full-text and hybrid vector search, natural-language questions and a visual query builder. | Search |
| AI assistant | Summaries, question answering, document comparison, grammar and rewriting on any document. | AI features |
| Bring your own model | Choose Claude, OpenAI, Gemini, DeepSeek, Ollama, vLLM or Ollama Cloud models per AI function. | AI features |
| Organization control | Invitations, access roles, departments and per-user menu exceptions. | Access control |
| Transparent credits | Pay-as-you-go credits with Stripe top-ups, auto charge and a full transaction history. | Billing |
Glossary#
| Organization (tenant) | Your isolated workspace. Documents, users, settings, credits and search indexes never cross organizations. |
|---|---|
| Document | One uploaded file (or one piece of a split file) with its metadata, text, extracted JSON and version number. |
| Content type | A file extension the platform accepts, such as .pdf, and the OCR engine used to read it. |
| Document type | A business category such as General Invoice. It owns the JSON schema used for extraction and its own search index. |
| JSON schema (template) | The fields to extract for a document type. |
| Status | Where a document is in the pipeline, from Waiting for pre processing to Searchable. |
| AI job | One request sent to the AI service (split, map JSON, search, compare, …) with its status and response. |
| Credit | The unit AI processing and storage are billed in. |
| System task | A scheduled background job that moves documents through the pipeline. |
Getting help#
- Press Ctrl + K to search these docs.
- Start with Troubleshooting & FAQ when something doesn't look right.
- Platform administrators can check Platform operations for task schedules, dashboards and logs.