Fluent AI Document Intelligence

Fluent AI is FluentCase’s built-in document intelligence engine. When a PDF lands in your Unsorted documents queue, Fluent AI reads its full text, identifies what the document is, links it to the right case matter, applies the correct tags, renames it according to your naming convention, and—if you have notification rules configured—alerts the appropriate staff.

The result: documents that used to sit in an inbox until someone had time to sort them are automatically filed, labeled, and routed in seconds.

Premium feature. Fluent AI processing is available to firms with AI-enabled plans.

How It Works

Every AI-processed document travels through the same pipeline. Each step hands its results to the next:

Upload

PDF lands in Unsorted

OCR

Text extracted from every page

AI Analysis

Document classified and summarized

Match

Document linked to a case matter

Tag

Tags applied by rules and AI

Rename

Filename built from your template

Notify

Alerts sent to matched recipients

OCR runs first regardless of your AI settings—it makes documents searchable and is the foundation Fluent AI builds on. Once OCR has extracted the text, the AI analysis step classifies the document type, generates a summary, and proposes a filename and tags. The match, tag, rename, and notify steps all run server-side in FluentCase before the result appears in your matter.

Enabling Fluent AI

Go to Settings → Fluent AI to configure AI processing for your firm. The settings panel is divided into two modes:

Link Only Faster

Auto-link unsorted documents to matters scans the first few pages of each incoming PDF to identify a matching case matter. The document is filed, and an AI tag is applied so staff know it was placed automatically. No summary is generated and no detailed classification occurs—this mode is fast and uses minimal resources.

Use this mode when your primary goal is routing: getting documents into the right matter quickly.

Summarize, Name, and Tag Full analysis

Summarize, name, and tag documents using AI performs full-document analysis. The entire contents of each PDF are read, and the AI produces:

A plain-language summary of the document
A document type classification from your firm’s taxonomy
Suggested tags based on content
A structured filename built from your naming template

Enabling this mode automatically enables link-only processing. This mode takes more time per document but delivers significantly richer results.

Batch Scan Auto-Detection

Automatically process batch scans if barcodes are detected monitors every incoming PDF for separator sheets. When a separator is found, FluentCase splits the scan into individual documents. Each resulting document is then processed through the full Fluent AI pipeline described above—OCR, AI analysis, matter matching, tagging, renaming, and notification—exactly as if it had been uploaded individually. This is especially useful for firms that scan large stacks of mixed documents at once. See Batch Scanning for instructions on preparing and uploading batch scans.

Document Naming Preferences

When Summarize, Name, and Tag is enabled, you can define exactly how Fluent AI builds filenames.

Prefix

A fixed prefix is prepended to every AI-generated filename, regardless of other naming settings. It is never altered by the case style setting.

Setting	Example
Prefix: `XSD`, Separator: `_`	`XSD_08-02-2026_Medical_Records.pdf`
Prefix: `AI`, Separator: `-`	`AI-08-02-2026-Medical-Records.pdf`
No prefix	`08-02-2026_Medical_Records.pdf`

Filename Components

Choose which components appear in the filename and drag them into your preferred order. Available components:

Component	Example value
Document Type	`Medical Records`
Matter	`Smith v. ABC Corp`
Date	`08-02-2026`
Claim Number	`WC-2024-5678`
Court Case Number	`CV-2024-9012`
Sender	`State Farm Insurance`
Subject / Re:	`Settlement Demand`

Delimiter

The character string placed between each component. Common choices: ` - ` (space-dash-space), _ (underscore), ` ` (space).

Date Format

Controls how dates appear in the filename. Examples:

08-02-2026 (MM-DD-YYYY with dashes)
August 2, 2026 (full month name)
2026-08-02 (ISO format)

Matter Format

When the Matter component is included, choose which part of the matter appears:

Name only — Smith v. ABC Corp
File number only — 2024-1234
Name (File Number) — Smith v. ABC Corp (2024-1234)
File Number – Name — 2024-1234 – Smith v. ABC Corp

Case Style

Controls capitalization of the assembled filename (the prefix is exempt):

Preserve — keep casing as returned by the AI
Title Case — Capitalize Each Word
UPPERCASE — ALL CAPS
lowercase — all lowercase

Preserve Manually Named Files

When Skip files that appear to have been manually labeled is checked, Fluent AI will not rename a file whose name already looks descriptive—for example, a file a staff member explicitly named before uploading. The static prefix is still applied even when the body of the filename is preserved.

Document Types

The Document Types panel shows your firm’s document type taxonomy. When the AI classifies an incoming document, it matches it against this list. The matched type can:

Drive the filename — the Document Type component in the filename comes from the matched type’s name.
Apply tags automatically — each document type has associated tags that are applied when a document is classified as that type.
Override the naming template — individual types can define their own field order and separators, letting you produce type-specific filename formats (for example, medical records might include the provider name, while legal notices include the court case number).

Types are evaluated in the order shown. You can reorder them using the up/down controls. Types can be enabled or disabled individually.

Deterministic Tag Rules

Deterministic rules let you tag documents based on their text content—without waiting for AI classification. Rules run in the order listed and can be used alongside or independently of the AI model.

How Rules Work

Each rule has a pattern (a word or phrase to look for) and a set of tags to apply when that pattern is found in the document text. Rules can also be assigned a document type, which causes the matched type to be used for naming even when the AI would have chosen differently.

Rule Priority and Stop Processing

Rules are evaluated from top to bottom. If a rule has Stop Processing enabled, evaluation halts as soon as that rule matches—no further rules are checked. This lets you create high-confidence, high-priority rules that take precedence over the rest.

Rules you import from your document type taxonomy are automatically placed at the end of the list with Stop Processing enabled, so any hand-crafted rules you place above them still take priority.

Match Types

Type	Behavior
Contains	Matches if the pattern appears anywhere in the document text (case-insensitive by default)
Regular Expression	Matches if the pattern is a valid regex that matches the document text
Conditions	Combines multiple terms with AND/OR logic for precise compound matching

Example Rules

Pattern: utilization review → Tags: UR, IMR
Pattern: notice of hearing → Tags: Hearing, RUSH
Pattern: QME report → Tags: QME, Document Type: QME Report

Allowed Tags

The Allowed Tags panel controls which tags the AI is permitted to suggest. This does not affect:

Tags applied by deterministic rules
Tags you apply manually
System tags like Document and AI

No restriction (default): The AI may suggest any tag in your firm’s tag library.

Restricted list: Only tags on the allowed list are eligible for AI suggestion. If the AI suggests a tag not on the list, that tag is silently dropped. This is useful for keeping AI tagging focused on a curated vocabulary—especially important when your firm has many tags.

To configure the list, click Edit in the Allowed Tags panel and check the tags you want to permit.

Document Notifications

Notification rules let you alert specific people whenever Fluent AI processes a document that matches certain tag criteria.

Notification Types

Type	What happens
Memo	A memo is created in the recipient’s FluentCase inbox with the document attached and the AI summary included in the body
Email	An email is sent to the recipient’s address with a document preview

Conditions

Condition	Behavior
Always	Notification fires for every AI-processed document
Any of [tags]	Notification fires if at least one of the listed tags was applied
All of [tags]	Notification fires only if every listed tag was applied

Recipients

Recipients can be named individuals or staff roles:

Attorney Responsible — the lead attorney on the matter
Attorney Handling — the handling attorney
Paralegal Handling — the assigned paralegal
Specific User — any active firm member

Example Notification Rules

Any document tagged RUSH → Send memo to Managing Attorney
Any document tagged Hearing → Send memo to Paralegal Handling
Document tagged both QME and Defense → Send email to outside counsel

How Tags Are Applied

Multiple systems can contribute tags to a single document. They are merged before being written to the activity:

Deterministic rules — evaluated first, in order; may stop early if a rule has Stop Processing enabled.
AI-suggested tags — returned by the AI model and filtered through the Allowed Tags list.
Document type tags — applied when the document matches a type in your taxonomy.
System tags — Document is always applied. AI is applied when the Apply AI tag setting is enabled (on by default).

Tags from different sources are combined, not replaced. A document can receive tags from all four systems simultaneously.

Tips

Start with Link Only, confirm matters are matching correctly, then enable Summarize for richer results.
Build your document type taxonomy first. The more precise your types, the more consistent your filenames.
Use deterministic rules for high-certainty patterns. If a phrase appears exclusively in one document type (like PR-4 or EAMS filing), a rule will catch it faster and more reliably than the AI alone.
Use the Allowed Tags list to prevent the AI from applying tags that don’t fit your workflow vocabulary.
Test notifications by processing a sample document and verifying the memo appears in the expected inbox before rolling out to the full team.