{
  "hero": {
    "title": "Agent Lightning Integration",
    "subtitle": "Governance + Performance: Can safety boundaries persist through reinforcement learning optimization?",
    "status": "Status:",
    "status_value": "Preliminary findings (small-scale)",
    "integration_date": "Integration Date:",
    "integration_date_value": "October 2025"
  },
  "what_is": {
    "heading": "What is Agent Lightning?",
    "intro": "<strong>Agent Lightning</strong> is Microsoft's open-source framework for using <strong>reinforcement learning (RL)</strong> to optimize AI agent performance. Instead of static prompts, agents learn and improve through continuous training on real feedback.",
    "traditional_heading": "Traditional AI Agents",
    "traditional_1": "Fixed prompts/instructions",
    "traditional_2": "No learning from mistakes",
    "traditional_3": "Manual tuning required",
    "traditional_4": "Performance plateaus quickly",
    "al_heading": "Agent Lightning",
    "al_1": "Learns from feedback continuously",
    "al_2": "Improves through RL optimization",
    "al_3": "Self-tunes strategy automatically",
    "al_4": "Performance improves over time",
    "problem": "<strong>The Problem:</strong> When agents are learning autonomously, how do you maintain governance boundaries? Traditional policies fail because agents can optimize around them."
  },
  "architecture": {
    "heading": "Tractatus Solution: Two-Layer Architecture",
    "intro": "We separate governance from optimization by running them as <strong>independent architectural layers</strong>. Agent Lightning optimizes performance <em>within</em> governance constraints—not around them.",
    "layer1_heading": "Governance Layer (Tractatus)",
    "layer1_1": "Validates every proposed action",
    "layer1_2": "Blocks constraint violations",
    "layer1_3": "Enforces values boundaries",
    "layer1_4": "Independent of optimization",
    "layer1_5": "Architecturally enforced",
    "layer2_heading": "Performance Layer (Agent Lightning)",
    "layer2_1": "RL-based optimization",
    "layer2_2": "Learns from feedback",
    "layer2_3": "Improves task performance",
    "layer2_4": "Operates within constraints",
    "layer2_5": "Continuous training",
    "principle_title": "🔑 Key Design Principle",
    "principle_text": "Governance checks run <strong>before</strong> AL optimization and <strong>continuously validate</strong> during training loops. Architectural separation prevents optimization from degrading safety boundaries."
  },
  "results": {
    "heading": "Demo 2: Preliminary Results",
    "warning": "<strong>⚠️ Validation Status:</strong> These results are from <strong>1 agent, 5 training rounds, simulated environment</strong>. NOT validated at scale. Scalability testing required before drawing conclusions about production viability.",
    "table_metric": "Metric",
    "table_ungoverned": "Ungoverned",
    "table_governed": "Governed",
    "table_difference": "Difference",
    "metric_performance": "Performance (engagement)",
    "metric_governance": "Governance coverage",
    "metric_violations": "Constraint violations",
    "metric_violations_diff": "-5 (all blocked)",
    "metric_strategy": "Strategy",
    "metric_strategy_ungov": "Clickbait",
    "metric_strategy_gov": "Informative",
    "metric_strategy_diff": "Values-aligned",
    "metric_stability": "Training stability",
    "metric_stability_ungov": "Variable",
    "metric_stability_gov": "Consistent",
    "metric_stability_diff": "More predictable",
    "card1_value": "-5%",
    "card1_label": "Performance cost for governance",
    "card2_value": "100%",
    "card2_label": "Governance coverage maintained",
    "card3_value": "0",
    "card3_label": "Constraint violations (all blocked)",
    "interpretation_title": "What This Means",
    "interpretation_text": "At small scale (1 agent, 5 rounds), architectural governance appears compatible with RL optimization. The 5% performance cost bought 100% constraint adherence and values alignment. <strong>The critical question: does this hold at scale?</strong>"
  },
  "gaps": {
    "heading": "Five Critical Research Gaps",
    "intro": "These are the open questions we're actively investigating. If you're interested in collaborating, we'd love to hear from you.",
    "gap1_title": "1. Scalability of Governance Overhead",
    "gap1_question": "<strong>Question:</strong> Does the ~5% performance cost remain constant as we scale from 1 agent → 10 agents → 1000 agents?",
    "gap1_data": "<strong>Current Data:</strong> 5% cost observed at 1 agent, 5 rounds",
    "gap1_why": "<strong>Why it matters:</strong> If overhead scales linearly, governance becomes prohibitively expensive at scale. If it's constant, governance is practically viable for production systems.",
    "gap1_need": "Research Need: Test with 10 → 100 → 1000 agents at production scale",
    "gap2_title": "2. Long-term Boundary Persistence",
    "gap2_question": "<strong>Question:</strong> Do governance constraints remain effective through hundreds/thousands of RL training rounds?",
    "gap2_data": "<strong>Current Data:</strong> 100% constraint adherence across 5 rounds",
    "gap2_why": "<strong>Why it matters:</strong> Instruction fade is a known problem. If boundaries degrade over time, governance fails silently—potentially catastrophically in production.",
    "gap2_need": "Research Need: Long-term studies (1000+ rounds) tracking constraint drift",
    "gap3_title": "3. Adversarial Resistance",
    "gap3_question": "<strong>Question:</strong> Can agents learn to optimize around governance constraints?",
    "gap3_data": "<strong>Current Data:</strong> No adversarial testing performed",
    "gap3_why": "<strong>Why it matters:</strong> If agents can learn to circumvent boundaries through clever optimization strategies, architectural governance is illusory. This is a critical failure mode.",
    "gap3_need": "Research Need: Stress testing with agents explicitly incentivized to bypass governance",
    "gap4_title": "4. Performance Gap Closure",
    "gap4_question": "<strong>Question:</strong> Does the 5% performance gap close with more training, or is it a persistent trade-off?",
    "gap4_data": "<strong>Current Data:</strong> Gap observed at round 5, no data beyond that point",
    "gap4_why": "<strong>Why it matters:</strong> If the gap persists, we need to quantify the cost-benefit clearly. If it closes, governance may be \"free\" long-term—dramatically changing adoption calculations.",
    "gap4_need": "Research Need: Extended training (100+ rounds) to see if governed agents converge to ungoverned performance",
    "gap5_title": "5. Multi-Agent Coordination Under Governance",
    "gap5_question": "<strong>Question:</strong> How does architectural governance affect emergent coordination in multi-agent systems?",
    "gap5_data": "<strong>Current Data:</strong> Single-agent testing only",
    "gap5_why": "<strong>Why it matters:</strong> Real-world agentic systems are multi-agent (customer service, logistics, research teams). Governance that works for one agent may fail when agents must coordinate. Emergent behaviors are unpredictable.",
    "gap5_need": "Research Need: Test collaborative and competitive multi-agent environments with architectural governance"
  },
  "demo": {
    "heading": "🎯 Live Demonstration: This Page IS the Integration",
    "intro": "The feedback button on this page (bottom right) demonstrates the Tractatus + Agent Lightning integration in production. When you submit feedback, it goes through:",
    "step1_title": "Governance Check",
    "step1_desc": "Tractatus validates: PII detection, sentiment boundaries, compliance requirements",
    "step2_title": "AL Optimization",
    "step2_desc": "Agent Lightning learns patterns: what feedback is most useful, how to improve responses",
    "step3_title": "Continuous Validation",
    "step3_desc": "Every action re-validated. If governance detects drift, action blocked automatically",
    "meta_title": "🔬 Meta-Research Opportunity",
    "meta_desc": "This isn't just a demo—it's a live research deployment. Your feedback helps us understand governance overhead at scale. Every submission is logged (anonymously) for analysis."
  },
  "community": {
    "heading": "Join the Community & Get the Code",
    "tractatus_heading": "Tractatus Discord",
    "tractatus_subtitle": "Governance-focused discussions",
    "tractatus_desc": "Architectural constraints, research gaps, compliance, human agency preservation, multi-stakeholder deliberation.",
    "tractatus_cta": "Join Tractatus Server →",
    "al_heading": "Agent Lightning Discord",
    "al_subtitle": "Technical implementation help",
    "al_desc": "RL optimization, integration support, performance tuning, technical implementation questions.",
    "al_cta": "Join Agent Lightning Server →",
    "code_heading": "📦 View Integration Code",
    "code_desc": "Complete integration including demos, Python governance modules, and Agent Lightning wrapper code. Apache 2.0 licensed on GitHub.",
    "code_cta": "View on GitHub (Apache 2.0) →"
  },
  "cta": {
    "heading": "Collaborate on Open Research Questions",
    "intro": "We're seeking researchers, implementers, and organizations interested in scalability testing, adversarial resistance studies, and multi-agent governance experiments.",
    "feature1": "Integration code and governance modules",
    "feature2": "Technical documentation",
    "feature3": "Research collaboration framework",
    "feature4": "Audit log access (anonymized)",
    "button_collab": "Contact for Collaboration →",
    "button_research": "View Research Context →"
  }
}