{
  "version": "2026-09-20",
  "status": "pre-launch-reviewed-facts-only",
  "canonicalPressUrl": "https://www.tryvira.xyz/press",
  "productUrl": "https://www.tryvira.xyz/product",
  "contactPath": "/contact",
  "oneLiner": "Vira Behavior is a control plane for teaching, evaluating, releasing, explaining, and rolling back governed AI behavior across models and applications.",
  "description50": "Vira Behavior gives AI behavior its own governed lifecycle. Teams can turn real corrections into reviewed behavior knowledge, evaluate exact candidates, release exact revisions through shadow and canary stages, explain runtime authority with receipts, and roll back deliberately—without treating the latest prompt or an optimizer as the final policy source.",
  "description150": "Vira Behavior is a behavior control plane for AI systems. It turns real corrections and expert judgment into reviewed behavior knowledge, then keeps evaluation, release, runtime explanation, optimization, and rollback attached to exact identities. Scope composition can separate project policy, application behavior, workspace configuration, user preference, and session state while protecting higher-authority hard rules. Behavior Points make intervention locations such as generation, decisions, tool actions, handoffs, verification, and escalation explicit. Candidates can be evaluated against exact baselines before shadow or canary release, and promotion remains an explicit governed action. Runtime receipts connect served behavior to the revisions, signals, projections, and model target that actually executed. Model-specific optimization may improve execution, but it does not become policy authority; stale or mismatched artifacts fall back to the canonical governed path. Vira is designed to sit around existing models, agent frameworks, and applications rather than replace them.",
  "architectureSummary": "Evidence → Lesson/candidate → exact evaluation → governed release → Behavior Point resolution → runtime receipt, with canonical behavior authority kept separate from model-target optimization.",
  "companyFacts": [
    {
      "label": "Product",
      "value": "Vira Behavior"
    },
    {
      "label": "Category",
      "value": "AI behavior control plane"
    },
    {
      "label": "Public site",
      "value": "https://www.tryvira.xyz/"
    },
    {
      "label": "Product URL",
      "value": "https://www.tryvira.xyz/product"
    },
    {
      "label": "Docs",
      "value": "https://docs.tryvira.xyz"
    },
    {
      "label": "Studio",
      "value": "https://app.tryvira.xyz"
    }
  ],
  "assets": {
    "brand": [
      {
        "label": "Vira brand icon — SVG",
        "href": "https://www.tryvira.xyz/icon.svg",
        "format": "SVG",
        "usage": "Preferred vector mark for press/editorial layouts."
      },
      {
        "label": "Vira app icon — PNG",
        "href": "https://www.tryvira.xyz/apple-icon",
        "format": "PNG",
        "usage": "Generated square app icon; not a replacement for a reviewed full wordmark asset."
      }
    ],
    "productPreviews": [
      {
        "label": "Behavior Point authority map",
        "href": "https://www.tryvira.xyz/product/behavior-point-map.svg",
        "note": "Curated product preview using non-customer example data; not a production customer screenshot."
      },
      {
        "label": "AI System readiness",
        "href": "https://www.tryvira.xyz/product/ai-system-readiness.svg",
        "note": "Curated product preview using non-customer example data; not a production customer screenshot."
      },
      {
        "label": "Runtime trace explainability",
        "href": "https://www.tryvira.xyz/product/trace-explainability.svg",
        "note": "Curated product preview using non-customer example data; not a production customer screenshot."
      }
    ]
  },
  "pressAngles": [
    {
      "label": "Technical",
      "text": "AI behavior should have exact revisions, evaluations, releases, explanation, and rollback—not only prompt versions."
    },
    {
      "label": "Enterprise",
      "text": "Vira separates company policy, application behavior, workspace configuration, user preference, and session context with explicit authority boundaries."
    },
    {
      "label": "Agentic AI",
      "text": "Agents can act; teams need explicit behavior authority in addition to traces that explain what happened."
    },
    {
      "label": "Portability",
      "text": "Canonical behavior authority can remain stable while the underlying model or model-target projection changes."
    }
  ],
  "availability": {
    "founderBio": {
      "status": "pending_review",
      "value": null,
      "note": "Publish only from reviewed founder/company facts; do not infer a biography from repository history."
    },
    "directMediaEmail": {
      "status": "pending_monitored_address",
      "value": null,
      "note": "Use the public contact route until a monitored media inbox is reviewed and staffed."
    },
    "launchDate": {
      "status": "not_announced",
      "value": null,
      "note": "Do not manufacture a public launch date from development timestamps."
    },
    "demoVideo": {
      "status": "pending_reviewed_release_capture",
      "value": null,
      "note": "Capture from a reviewed release candidate after browser QA."
    },
    "realProductScreenshots": {
      "status": "pending_reviewed_release_capture",
      "value": null,
      "note": "Current assets are curated non-customer product previews, not claimed customer screenshots."
    }
  },
  "faq": [
    {
      "question": "What problem does Vira Behavior address?",
      "answer": "It gives AI behavior changes an explicit lifecycle for evidence, exact revisions, evaluation, release, runtime explanation, and rollback."
    },
    {
      "question": "Does Vira replace a model provider or agent framework?",
      "answer": "No. Vira is designed to add a governed behavior authority layer around existing models, frameworks, and applications."
    },
    {
      "question": "Does optimization become policy authority?",
      "answer": "No. Model-target optimization remains attached to exact governed behavior identity; stale or mismatched artifacts fall back to the canonical path."
    },
    {
      "question": "Are public benchmark claims available?",
      "answer": "Not unless a dated first-party methodology, exact model and dataset identity, sample size, results, and limitations are published."
    }
  ]
}