Tag Features
Everything you need for AI-assisted document workflows—from data collection through professional output
Smart Data Collection
Traditional data collection is tedious: open a document, find the relevant information, switch to your form, paste it in, repeat 50 times. Tag transforms this into a streamlined workflow.
Batch Autofill Wizard
Select dozens of source documents for autofill at once from anywhere on your computer. The wizard provides:
Multi-folder selection - Pick files from different locations
Favorite folders - Save frequently used source folders for quick access
Document-to-form mapping - Attach relevant source documents to each form that needs data, process them separately or together in AI requests
Per-form AI configuration - Choose AI vendor and model for each form independently
Routing flexibility - Send requests to a secure AWS Bedrock server (Managed AI subscription) or to the AI vendor’s server directly (BYOK subscription) per form
Parallel processing - Up to 30 source documents process simultaneously, more than that safely queue for their turn
Real-time progress - Progress bars show what's running; detailed JSON traces show what happened
How it works: You design forms with fields that match the data you need. Data fields use public standards (RNG, XML Schema) to define structure and validation rules. When you run batch autofill, AI scans your source documents, finds the specific information for each field, and populates the form.
Tag sends a big system prompt to explain how this works to the AI. You can enhance this by attaching a custom prompt, or typing a custom instruction at runtime (e.g., Ignore all events prior to 2023, only fill fields A, B and C).
When autofill is done, you review and correct data before it is transformed and/or flowed into final documents. Autofill can be run multiple times if you need more control over the sequence of source documents used.
Advanced Data Import
Unlike simple form builders that only accept manual typing, Tag pulls data from multiple sources:
Excel/CSV imports with field-level transformations (e.g., reformat dates, split names, apply calculations)
Advanced find helps you search for a single row in a spreadsheet to import from
Transformation rules can change values (e.g., from a survey or online Google Form) into exactly what your data fields expect
Database queries (SQL: MySQL, PostgreSQL, SQL Server, Oracle; NoSQL: MongoDB, CouchDB)
API calls to web services and cloud systems
Semantic data endpoints for knowledge graphs and linked data
Some data sources require a bit of help from us to setup. Ensuring user credentials are safely managed is of utmost importance.
Targeted AI Processing
The difference between AI that helps and AI that frustrates is control. Tag doesn't generate entire documents from vague prompts - it processes specific data with specific instructions.
AI-Chains for Precision
AI-Chains bundle everything needed for a focused task: the data to process, the context to provide, the prompt that instructs the AI, and model settings.
How it works:
Define data sources - Select which form fields, local files, or other data the chain sees
Connect prompts - Pull from your prompt library or write inline
Choose vendor and model - Use Claude for complex reasoning, Gemini for fast extraction, or compare multiple models and decide for yourself
Attach to form fields - By convention, fields named *_ai support AI-Chain attachment
Execute on demand - Run on one field, one form, or all forms together
What makes this powerful:
Granular control - AI processes exactly the data you specify, nothing more
Reusable patterns - Save successful chains for similar tasks elsewhere
Vendor flexibility - Switch between Anthropic, Google, OpenAI, or Cohere per task (some vendors are not available in our Managed AI subscription)
Complete transparency - Detailed JSON traces show every request and response (required for HIPAA/PIPEDA audit trails)
No hidden prompts - Tag system prompts are visible in the trace log; everything else is in your prompt library or explicitly provided by you
Writing Samples Inform Style
Instead of relying on AI's default voice, provide samples of your actual writing. AI analyzes the examples and matches:
Tone and formality level
Sentence structure and paragraph length
Technical vocabulary appropriate to your field
Organizational patterns you consistently use
This isn't just prompt engineering - it's teaching AI to write like you.
Writing samples are key to getting good results. Our experience with healthcare professionals has shown that different authors typically want completely different output content for the same set of input data. As a result, customizing writing samples is an important part of getting started.
Professional Document Generation
Tag excels at generating complex documents that maintain your standards and branding. The template designer gives complete control without requiring coding expertise.
Rich Template Designer
Create sophisticated layouts with full formatting control:
Tables with dynamic rows, merged cells, calculated totals
Sections with conditional inclusion based on data
Styles and formatting - fonts, colors, spacing, headers/footers precisely as you want them
Mixed content - combine fixed text (disclaimers, standard language) with dynamic fields and AI-generated sections
Repeating elements - generate tables or lists of items in your data
Nested structures - complex documents with multiple levels of organization
Powerful Expression Language
XPath-based expressions - leverage the full power of XPath functions for data manipulation and logic
Conditional content - show or hide sections, paragraphs, or table rows based on expression results
Dynamic repetition - loop through data collections to generate tables, lists, or document sections
Data insertion - pull values from your data source using value-of expressions
Transform on import - apply expressions during data import to calculate, format, or restructure information
Point-and-click editors - build common expressions visually without writing code
Raw expression editing - drop into code for advanced scenarios requiring fine-tuned control
Flexible Integration
Real-world workflows involve data from many places in many formats. Tag handles integration complexity so you don't have to.
Desktop Architecture Advantages
Tag runs on your computer (Windows or Mac), not in a browser:
Performance benefits:
Fast processing of large files without upload/download delays
Parallel operations leverage your CPU cores
Reduced impact of variable server availability or network congestion
Utilize larger amounts of local memory instead of a slice of remote server memory
Security advantages:
Files stored locally, not on remote servers
You control what data leaves your computer
Connect to databases on private networks (VPN, corporate intranet)
Access file shares behind firewalls
Work with systems that can't be exposed to the public internet
XML and JSON Capability
Tag has flexible support for structured data:
XML parsing and generation with schema validation
JSON handling for modern web APIs
Bidirectional conversion - read XML, write JSON; read JSON, write XML
XSLT transformations for complex document generation
XPath and JSONPath queries for precise data extraction
This makes Tag suitable for industries with strict data interchange requirements (healthcare HL7, financial FIXML, government standards).
What You Won't Find in Tag
Autonomous AI agents: Tag does not include AI agents that independently take actions. Every AI operation is explicitly triggered by a human clicking a button. This is deliberate - we believe in keeping humans in control.
Generic document generation: Unlike tools that generate entire documents from a single vague prompt, Tag requires you to design workflows. This takes more initial setup but produces predictable, highly customized results.
Cloud-based storage: Tag is a desktop application. If you need cloud storage for your files, use your existing solutions (Google Drive, OneDrive, Dropbox). Tag reads from and writes to wherever you store files locally.
Real-time collaboration editing: Multiple users can contribute to shared forms asynchronously, but Tag doesn't support Google Docs-style simultaneous editing of documents. Final document generation happens on one person's computer.
These are intentional design decisions aligned with our philosophy: human control, predictable workflows, security, and compliance.
Tag is available for Windows and macOS. All features described are included in both Individual-BYOK and Individual-Managed subscription plans. Blueprint modeling features are available through consulting engagements.

