AI & Knowledge Management
Where Human Expertise Meets Machine Intelligence
V.E.T.S. is building the world's most comprehensive animal care knowledge base. Our AI integration doesn't replace expert judgment—it amplifies it. AI generates initial documentation to bootstrap engagement, domain experts refine through daily use, and the system captures collective wisdom as a natural byproduct of real work.
The Vision: Every procedure, every protocol, every piece of institutional knowledge—captured, organized, and continuously improved. Not through burdensome documentation requirements, but through intelligent systems that learn from how experts actually work.
AI Integration: Live & Operational
As of November 2025, our AI integration is fully deployed and operational in the production environment.
Generate Description
One click generates comprehensive descriptions for procedures, medications, equipment, and tree value categories. AI understands context—whether you're documenting a surgical technique or organizing a service category.
TeamDoc Sections
Automatically generates structured documentation sections: Indications, Procedures, Contraindications, Follow-up Care, Equipment Needed, and more. Templates adapt based on what you're documenting.
AI Classification
Machine-readable rules help future AI understand your data. JSON-formatted guidelines capture what belongs where, edge cases, and classification criteria—making the system smarter over time.
Content Review
AI reviews and enhances existing documentation, suggesting improvements, identifying gaps, and ensuring consistency across your knowledge base.
The Knowledge Bootstrap Problem
Every knowledge system faces the same impossible challenge:
The Chicken
You need content to attract experts
The Egg
You need experts to create content
The Result
Empty systems that nobody uses
Traditional Approaches Failed Because:
- Empty knowledge bases provide no immediate value
- Busy professionals can't document everything from scratch
- Documentation becomes a burden separate from real work
- Systems fill with stale, unmaintained content
The V.E.T.S. Solution: AI generates comprehensive initial documentation for every procedure, medication, and technique. This gives domain experts something concrete to work with immediately—instant value from day one, without asking anyone to start from a blank page.
How AI Content Generation Works
The Workflow
- Context Capture: When you click the AI icon, the system gathers everything relevant—item name, category, species, parent nodes, related procedures, existing documentation.
- Intelligent Prompting: Your context is sent to the AI with carefully engineered prompts. The AI knows whether you're documenting a surgical procedure, a medication, equipment, or a category hierarchy.
- Structured Generation: AI returns comprehensive content organized into meaningful sections—not a wall of text, but structured documentation ready for expert review.
- Review & Approve: You see exactly what AI generated. Check the sections you want to keep, edit what needs adjustment, discard what doesn't fit.
- Save & Build: Approved content creates TeamDoc sections automatically. Your knowledge base grows with one click.
Intelligence Features
Tree Value Awareness: The AI understands when you're documenting an organizational category ("Equine Surgical Procedures") versus a specific procedure ("Cryptorchid Castration"). Different contexts get different content.
TransactionType Templates: Section templates adapt based on what you're documenting. Surgical procedures get different sections than medications, which get different sections than equipment.
The Knowledge Capture Loop
This is where V.E.T.S. fundamentally differs from traditional knowledge systems.
1. Generate
AI creates initial documentation based on item context and domain knowledge.
2. Use
Experts use the system daily for actual work—managing animals, documenting care, collaborating.
3. Refine
As experts work, they naturally correct, enhance, and add nuance to documentation.
4. Capture
Every refinement improves the canonical knowledge base—no separate documentation effort.
Refinement in Action
- Edit a procedure description → Updates canonical documentation for all users
- Add a dosing note → Captured in treatment protocols system-wide
- Link related procedures → Builds the knowledge graph automatically
- Document a complication → Enhances cautions and contraindications
- Correct an AI error → System learns what experts actually know
The Key Insight: Documentation becomes a byproduct of doing real work, not a separate burden. Every expert interaction makes the system smarter.
Flexible AI Architecture
V.E.T.S. doesn't lock you into a single AI provider. Our architecture is designed for flexibility as the AI landscape evolves.
Provider Interface Pattern
Swap AI providers without code changes. Today it's Gemini, tomorrow it could be Claude, GPT-5, or an on-premise local LLM. The architecture supports them all.
System Credentials
Organization-wide default AI provider. Everyone gets AI capabilities out of the box.
User Credentials
Power users can configure their own AI provider and API key. Use your organization's enterprise agreement, or experiment with different models.
Current Implementation
Default Provider: Google Gemini 2.5 Flash
Timeout: 120 seconds (for comprehensive content generation)
Features: Generate Description, Review Content, Section Templates, Classification Guidelines
Future-Proof: As AI models improve, V.E.T.S. can adopt them immediately. When Gemini 3 launches, or a better provider emerges, switching is a configuration change—not a rewrite.
AI-Enhanced Operations
Intelligence built into every layer—not bolted on, but integrated into how you already work.
Smart Documentation
AI assists with clinical notes, procedure documentation, and form completion. Context-aware suggestions based on species, history, and established protocols.
Knowledge Amplification
Your expertise grows the system. AI learns from your decisions, surfaces relevant history, and connects related records automatically. The more you use it, the smarter it gets.
Workflow Assistance
From reports to website content, AI helps draft, refine, and organize. Reduce the friction of administrative work so you can focus on animals.
The Friction Reduction Philosophy
The apps aren't broken. The workflow is just too heavy for daily use. AI reduces friction on systems that already work:
| Task | Current Friction | AI-Reduced Friction |
|---|
| Document a procedure | Start from scratch, remember format | AI drafts, you refine |
| Classify new items | Research, decide, organize manually | AI suggests, you confirm |
| Build knowledge base | Months of dedicated effort | Generate foundation in hours |
Field Intelligence
Ranch and farm operations generate massive amounts of data—feed consumption, animal movements, visual observations. V.E.T.S. captures it all and puts AI to work finding patterns you'd miss.
Feed & Nutrition AI
Track every pound fed. Set ration rules by species, sex, and condition. AI monitors consumption patterns, flags anomalies, and correlates feed data with weight gain and body condition scores.
Location Intelligence
Animal location tracking with movement history and timestamps. AI-assisted pasture rotation planning. Camera integration can auto-update locations—no manual entry required.
Visual AI
Connect your camera systems. AI identifies individual animals, scores body condition, monitors foaling mares, detects behavioral changes, and alerts you to problems before they become emergencies.
Infrastructure Ready: The database structures for feed tracking, location history, and inventory management already exist. The foundation is built. AI integration is the next layer.
Performance Video Analysis
For bucking horses, rodeo stock, and performance animals, V.E.T.S. goes beyond basic tracking into territory nobody else is touching.
The Workflow
- Film the buck-out or training session
- AI analyzes video for performance metrics
- Objective scores for kick, jump, intensity, pattern, rhythm
- Track over time—performance trending up or down?
- Correlate with data—training regimen, feed changes, conditioning work
- Actionable insights—what makes your top performers excel?
Traditional Judging
Two judges score horse and rider 1-25. Subjective, varies by judge, no breakdown of why.
AI-Enhanced Scoring
Objective metrics for kick power, jump height, intensity, pattern consistency, rhythm. Compare across sessions, identify trends.
Data Correlation
Connect performance to training logs, feed changes, rest periods. "Best scores correlate with 3-day rest" or "Performance dropped after feed change."
Competitive Advantage: No other platform combines video analysis + feed tracking + medical records + training logs + breeding data. This is a niche nobody else is touching.
AI Classification Guidelines
A unique feature that makes your knowledge base truly intelligent: machine-readable rules that teach AI how to organize and classify your data.
What Gets Captured
- Must Include: Keywords and criteria that define what belongs in a category
- Must Exclude: Items that seem like they might belong but don't—with explanations of where they actually go
- Edge Cases: Ambiguous situations and how to handle them
- Examples: Concrete items with reasoning for why they do or don't belong
Why This Matters
When you generate AI Classification Guidelines for a tree node, you're not just documenting—you're teaching. Future AI queries can read these rules to:
- Automatically suggest where new items should be categorized
- Flag potential misclassifications
- Build consistent organizational structure across your knowledge base
- Learn the subtle distinctions that make sense to your domain experts
JSON Format: Classification guidelines are stored as structured JSON—not prose. This makes them machine-readable, queryable, and usable by AI systems for automated organization.
AI Security & Accountability
AI capabilities are powerful—which means they need to be controlled, audited, and secure.
Permission Validation
Every AI request validates that you have permission to edit the item. The system checks both table-level permissions AND TeamDoc membership. No AI-generated changes to content you don't control.
Complete Audit Trail
Every AI request is logged: who made it, what they asked for, what was generated, how long it took, and how many tokens were used. Full accountability and cost tracking.
Human Review Required
AI generates suggestions—it never auto-commits content. Every piece of AI-generated content requires human review and explicit approval before it becomes part of your knowledge base.
What Gets Logged
- User login and timestamp
- AI provider and model used
- Context type (item vs. tree value)
- Full prompt sent to AI
- Full response received
- Token usage (for cost tracking)
- Response time
- Success/failure status
The Knowledge Feedback Loop
V.E.T.S. is designed to get smarter over time through multiple feedback mechanisms.
RAG
Retrieval-Augmented Generation searches existing TeamDoc content to ground AI responses in your expert-validated knowledge.
Fine-Tuning
Expert corrections create training data. Over time, models can be fine-tuned on your domain-specific knowledge.
Prompt Engineering
Validated content feeds into future prompts. AI learns from what experts have already approved.
Expert Corrections Drive Improvement
When you correct an AI-generated description, you're not just fixing one document—you're teaching the system:
- Validate: Mark content as expert-approved
- Correct: Fix errors, and the correction becomes training data
- Enhance: Add nuance ("in summer, follicles are 5mm smaller")
- Connect: Link to related procedures, creating knowledge graph edges
- Document exceptions: Capture the edge cases AI doesn't know
The Goal: Eventually, AI suggestions become so good that expert review is quick confirmation rather than substantial editing. But experts always remain in control.
Future AI Capabilities
The foundation is built. Here's where we're headed:
Intelligent Search
Natural language queries across your knowledge base. "Show me equine reproductive procedures requiring sedation" or "What protocols mention colic?"
Predictive Analytics
Suggest treatment pathways based on historical outcome data. Correlate interventions with results across your patient population.
Multi-Modal AI
Image analysis for radiographs, wound documentation, dermatology cases. Body condition scoring from photos. Video analysis for gait and behavior.
Suggest Field Values
AI suggests form field values based on context: PHDetails evaluations, SOAP components, treatment parameters. Reduce data entry to confirmation.
Infrastructure That Enables This
- Camera network: 30+ IP cameras with iSpy Agent integration ready
- Feed tracking: Database structures exist with historical data
- Location system: Animal movement tracking infrastructure in place
- Inventory: Schema ready with accounting integration hooks
- Provider flexibility: Swap AI providers as technology improves
For Developers: AI Integration Quick Reference
Key Database Objects
| Object | Purpose |
|---|
atbl_AI_SystemCredentials | System-wide API credentials (Provider, APIKey, Model) |
atbl_AI_UserCredentials | User-specific API credentials for power users |
atbl_AI_UsageLog | Audit trail for all AI requests |
atbl_AI_TeamDocSubjects | Section templates by TransactionType |
astp_AI_GetItemContext | Gathers comprehensive context for prompts |
astp_AI_CheckItemPermission | Validates user has edit access |
astp_AI_GetApplicableSections | Returns section templates for item type |
astp_AI_CreateTeamDocSection | Creates TeamDoc section from AI content |
Key Files
| File | Purpose |
|---|
AIAssistAJAX.aspx | AJAX endpoint for all AI requests |
AI.js | Client-side JavaScript (3,698 lines) |
ILLMProvider.vb | Provider interface for LLM flexibility |
Gemini25FlashProvider.vb | Current Gemini implementation |
Adding a New Provider: Implement ILLMProvider, update CreateProvider() in AIAssistAJAX, add credentials to atbl_AI_SystemCredentials. No other code changes required.
Build the Future of Animal Care Knowledge
AI doesn't replace expertise—it amplifies it. V.E.T.S. is building a platform where every veterinarian, breeder, trainer, and livestock professional contributes to a growing body of knowledge that benefits everyone.
For Veterinarians
Spend less time on documentation, more time on patients. Access collective wisdom from thousands of procedures. Let AI handle the routine so you can focus on the complex.
For Breeders & Trainers
Data-driven decisions backed by performance metrics, not just intuition. Correlate training, nutrition, and genetics with outcomes. Build competitive advantage through better information.
For Patients
Better outcomes from evidence-based protocols. Faster diagnosis through pattern recognition. Care informed by the collective experience of thousands of similar cases.
Get Started: AI features are live now. Click the AI icon next to any item or tree value to generate descriptions, build documentation sections, and start contributing to the knowledge base.