nSymbol Tag Overview
Professional Document Generation with AI-in-the-Loop
A desktop application for Windows and Mac that puts you in control of AI-assisted document workflows
Gathering Data - The Hard Part Made Easy
Traditional document generation starts with a painful process: manually entering data, chasing down information across multiple sources, copying and pasting from old files, and hoping you didn't miss anything critical.
Tag transforms data collection into a streamlined workflow.
Forms automatically adapt to your needs, pulling data from wherever it lives: uploaded documents (PDFs, Word files, scans), databases and APIs, Excel spreadsheets with field-level transformations, or manual entry when needed.
Where AI makes the difference:
Instead of reading a 20-page assessment report and manually typing key details into fields, you upload the document and let AI extract exactly what you need. Tag doesn't just dump all the text—it searches for specific information and maps it to the right fields.
You can chain multiple AI operations together: scan several documents, pull test scores from each, calculate summaries, and populate a comprehensive form—all in minutes instead of hours.
The staging area concept:
Forms become a verification point where you review what AI extracted before it flows into your final document. No hallucinations reach your output because you catch them here. When multiple team members contribute data, everyone works with the same shared forms, adding their pieces as information becomes available.
This is where Dashboard provides getting-started wizards, Scribe manages your forms, and Automate runs AI extraction with your choice of models (Claude, Gemini, OpenAI, or Cohere).
What Makes Tag Different
### No Black Box AI
Unlike tools that generate entire documents from a single prompt (with unpredictable results), Tag breaks complex tasks into smaller, focused operations. Each AI interaction has a specific job: extract these dates, summarize this section, classify this document type.
You see exactly what AI is doing because you design each step. Prompts aren't hidden—they're in your library, versioned and testable.
### Human-at-the-Wheel
AI assists, but humans decide. You choose where AI helps (data extraction, first drafts, pattern recognition) and where human judgment is required (interpretation, clinical decisions, final approval).
Every AI-assisted workflow has explicit review points. You're never locked into accepting AI suggestions—you verify, modify, or reject as appropriate.
### Your Data Stays Yours
Tag is a desktop application, not a web service. Your files live on your computer, not on our servers. When you use AI features, Tag connects directly to AI providers (like AWS Bedrock for our Managed plan, or your own API keys for BYOK plans) with encrypted requests that aren't retained for training.
This architecture enables HIPAA and PIPEDA alignment. We don't claim "Tag is HIPAA compliant" (compliance requires organizational policies beyond software), but the design supports regulatory requirements for healthcare, education, and other sensitive data environments.
Transforming Data - AI as Your Assistant
You've verified your data. Now you need to make sense of it—pull out patterns, generate summaries, create narrative content that sounds like you.
This is where AI-chains shine.
An AI-chain bundles everything needed for a specific task: the data to process, the context to provide, the prompt that instructs the AI, and writing samples that guide style. Save it once, run it with a click.
Reusable across similar tasks:
Created a chain that summarizes cognitive test results? Use the same pattern for academic tests, behavioral assessments, or speech-language evaluations. The prompt structure transfers—you just adjust the specific context and samples.
No vendor lock-in:
Switch between Anthropic Claude, Google Gemini, OpenAI GPT, or Cohere models effortlessly. Compare results. Use different vendors for different tasks based on speed, cost, or capability. When a model gets upgraded or a vendor changes pricing, you can adapt without rebuilding your workflows.
Full transparency:
Every AI interaction generates a detailed JSON trace showing exactly what was sent and received. For PIPEDA and HIPAA requirements, this creates the audit trail auditors expect. For quality improvement, it shows you what's working and what needs refinement.
No hallucinations on verified data:
Because AI-chains work on data you've already verified in the staging area, and because prompts are focused on specific tasks with clear context, the risk of AI making things up drops dramatically. You're not asking it to generate an entire report from memory—you're asking it to process verified information according to your instructions.
Automate manages your AI-chains library. Blueprint (available through consulting engagements) models complex workflows that combine multiple chains with human checkpoints and data routing logic.
Generating Documents - Professional Quality, Guaranteed
You have clean data. You've processed it with targeted AI assistance. Now you need a professional document that meets your standards.
Tag's template designer gives you complete control.
Design sophisticated layouts with tables, sections, conditional content, and custom styling—all without coding. Mix fixed content (your standard language, disclaimers, formatting) with dynamic sections that pull from your data or AI-generated text.
Multiple output formats:
Generate Word documents (.docx), Google Docs, or fillable PDFs. The same template can produce all three. Your formatting, branding, and professional standards remain consistent across formats.
Document catalogs for common needs:
Tag includes curated catalogs for psychological assessments (WAIS, WIAT, WISC, and 60+ other protocols), speech-language pathology evaluations, VA disability benefits questionnaires, and educational reports.
These aren't generic templates—they're built by subject-matter experts who understand the assessments. Data models, prompts, and writing samples are already configured. You just provide the source data.
For private or organizational needs, custom catalogs make your document types point-and-click accessible. Your team's expertise becomes institutional knowledge rather than staying in individuals' heads.
Aggregation and analysis:
Need to track trends across many documents? Tag can combine data from multiple completed documents into spreadsheets for analysis. See patterns across your caseload, generate reports for grant applications, or satisfy reporting requirements for funders.
Scribe handles all document generation. Dashboard provides catalog browsing and one-click access to installed document types.
Governance and Growth - Built for the Long Term
As your use of AI grows, management becomes critical. Which prompts work well? What happens when models change? How do you ensure team members use AI consistently and safely?
Tag is designed for mature, sustainable AI adoption.
Security by design:
Data processed locally on your desktop. Encrypted transit to AI services. You control what information gets sent to which AI vendors. Data masking features let you send fake data instead of real PII when testing or demonstrating workflows.
Machine-readable architecture:
Blueprint (available through consulting engagements) creates formal models of your workflows using OWL ontologies. This isn't just documentation—it's a queryable knowledge base. Ask questions like: "Which workflows use this AI model?" "Which prompts process protected health information?" "Where do humans perform reviews?"
Prompt libraries as organizational knowledge:
Successful prompts become reusable assets. Version them, test them, share them across your team. When models change or new team members join, institutional knowledge doesn't walk out the door.
Start simple, grow deliberately:
Most users begin with Dashboard for catalog-based documents or simple form-filling. They expand to Scribe for custom templates and Automate for their own AI-chains. Organizations concerned with governance and integration adopt Blueprint for formal workflow modeling.
This progression is intentional. You shouldn't need to understand enterprise architecture to generate your first report. But when you're ready to scale, the tools are there.
Compliance support:
Our Managed AI plan routes requests through AWS Bedrock with guarantees of zero data retention and no training on your data/prompts. For BYOK plans, you connect directly to vendors and follow their policies. Either way, audit trails capture every AI interaction, supporting HIPAA, PIPEDA, GDPR, and other regulatory frameworks.
Blueprint makes compliance visible: data flow diagrams show exactly where sensitive information goes, which models touch it, and where human oversight occurs.
---
## The Four Apps Working Together
Tag consists of four integrated applications, each with a specific role:
Dashboard - Your starting point for getting-started wizards, catalog browsing, and quick access to common tasks. Launch forms, run AI extraction, generate documents—all from one place.
Scribe - Template designer and document generator. Create rich templates with full formatting control. Design forms that collect data from multiple sources. Generate Word docs, Google Docs, or fillable PDFs.
Automate - AI-chain builder and executor. Create reusable AI operations that bundle context, prompts, and writing samples. Switch between AI vendors. Capture detailed audit trails. Run batch operations on up to 50 documents in parallel.
Blueprint - Workflow modeler and architecture visualizer (consulting-only). Create formal models of your data flows using OWL ontologies. Visualize where AI touches data, where humans review, how systems integrate. Query your architecture to support governance and change management.
You don't need all four apps for every task. Solo practitioners might work primarily in Dashboard and Scribe. Small teams might add Automate for custom prompts. Enterprises adopt Blueprint for formal governance.
The apps share data seamlessly. Forms created in Scribe feed AI-chains in Automate. Catalogs browsed in Dashboard generate documents in Scribe. Workflows modeled in Blueprint orchestrate across all three.