{
  "version": "https://jsonfeed.org/version/1.1",
  "title": "MedPAC Technologies — Insights",
  "home_page_url": "https://www.medpactech.com/blog",
  "feed_url": "https://www.medpactech.com/api/posts.json",
  "description": "Syndication feed for MedPAC Technologies product sites. Filter client-side on the products array.",
  "_medpac": {
    "products": [
      {
        "slug": "qurvia",
        "name": "Qurvia",
        "site": "https://www.qurvia.com"
      },
      {
        "slug": "medinsights",
        "name": "MedInsights",
        "site": "https://www.medinsights.com"
      },
      {
        "slug": "medrounds",
        "name": "MedRounds",
        "site": "https://www.medrounds.com"
      },
      {
        "slug": "hubbsly",
        "name": "Hubbsly",
        "site": "https://www.hubbsly.com"
      },
      {
        "slug": "ibeam",
        "name": "iBeam",
        "site": "https://www.ibeam.dev"
      },
      {
        "slug": "bindry",
        "name": "Bindry",
        "site": "https://www.bindry.ai"
      }
    ],
    "usage": "Filter items by the products array; always link to the canonical url."
  },
  "items": [
    {
      "id": "https://www.medpactech.com/blog/what-post-acute-teams-get-wrong-about-clinical-ai",
      "url": "https://www.medpactech.com/blog/what-post-acute-teams-get-wrong-about-clinical-ai",
      "title": "What post-acute teams get wrong about clinical AI",
      "summary": "Most clinical AI pilots in skilled nursing fail on workflow, not accuracy. What actually determines whether a model changes care.",
      "answer": "Clinical AI pilots in post-acute care usually fail because of workflow placement rather than model accuracy. A prediction that arrives outside the moment a clinician is already making a decision gets ignored regardless of how accurate it is. The three factors that determine whether clinical AI changes care are: whether the output appears inside an existing workflow step, whether the reasoning is traceable to source documentation, and whether a named clinical role owns acting on it.",
      "content_text": "Clinical AI pilots in post-acute care usually fail because of workflow placement rather than model accuracy. A prediction that arrives outside the moment a clinician is already making a decision gets ignored regardless of how accurate it is. The three factors that determine whether clinical AI changes care are: whether the output appears inside an existing workflow step, whether the reasoning is traceable to source documentation, and whether a named clinical role owns acting on it.",
      "date_published": "2026-09-02T12:00:00Z",
      "date_modified": "2026-09-02T12:00:00Z",
      "authors": [
        {
          "name": "Christian Armstrong, MD, MHSA"
        }
      ],
      "tags": [
        "Clinical AI"
      ],
      "products": [
        "qurvia",
        "medinsights"
      ],
      "reading_minutes": 7
    },
    {
      "id": "https://www.medpactech.com/blog/ai-agents-with-humans-orchestrating",
      "url": "https://www.medpactech.com/blog/ai-agents-with-humans-orchestrating",
      "title": "We build software with AI agents. Humans orchestrate.",
      "summary": "How MedPAC actually ships: AI agents do the typing, human staff direct the work, and every change is traceable from prompt to production.",
      "answer": "MedPAC Technologies develops software using AI agents directed by human staff rather than by engineers writing code line by line. Humans define the problem, set constraints, review at approval gates, and own the decision to ship; agents do the implementation work. The practices that make this safe are explicit approval gates before any merge, traceability from the originating prompt to the shipped change, and treating accumulated context as a versioned asset rather than something retyped each session.",
      "content_text": "MedPAC Technologies develops software using AI agents directed by human staff rather than by engineers writing code line by line. Humans define the problem, set constraints, review at approval gates, and own the decision to ship; agents do the implementation work. The practices that make this safe are explicit approval gates before any merge, traceability from the originating prompt to the shipped change, and treating accumulated context as a versioned asset rather than something retyped each session.",
      "date_published": "2026-08-19T12:00:00Z",
      "date_modified": "2026-08-19T12:00:00Z",
      "authors": [
        {
          "name": "Christian Armstrong, MD, MHSA"
        }
      ],
      "tags": [
        "AI Build"
      ],
      "products": [
        "ibeam",
        "bindry",
        "hubbsly"
      ],
      "reading_minutes": 9
    },
    {
      "id": "https://www.medpactech.com/blog/baa-questions-to-ask-an-ai-vendor",
      "url": "https://www.medpactech.com/blog/baa-questions-to-ask-an-ai-vendor",
      "title": "Eleven questions to ask an AI vendor before you sign the BAA",
      "summary": "A procurement checklist for healthcare organizations evaluating AI vendors, written from the vendor side of the table.",
      "answer": "Before signing a Business Associate Agreement with an AI vendor, a healthcare organization should establish: whether protected health information is used to train models, which subprocessors receive PHI, where data is hosted, how long data is retained after termination, whether outputs are traceable to source records, who is liable for a clinical decision influenced by the tool, and whether the vendor will support an audit. Get each answer in the contract rather than in a sales call.",
      "content_text": "Before signing a Business Associate Agreement with an AI vendor, a healthcare organization should establish: whether protected health information is used to train models, which subprocessors receive PHI, where data is hosted, how long data is retained after termination, whether outputs are traceable to source records, who is liable for a clinical decision influenced by the tool, and whether the vendor will support an audit. Get each answer in the contract rather than in a sales call.",
      "date_published": "2026-07-30T12:00:00Z",
      "date_modified": "2026-08-12T12:00:00Z",
      "authors": [
        {
          "name": "Christian Armstrong, MD, MHSA"
        }
      ],
      "tags": [
        "Compliance"
      ],
      "products": [
        "qurvia",
        "medinsights",
        "medrounds"
      ],
      "reading_minutes": 6
    },
    {
      "id": "https://www.medpactech.com/blog/readmission-math-nobody-runs",
      "url": "https://www.medpactech.com/blog/readmission-math-nobody-runs",
      "title": "The readmission math nobody runs before buying software",
      "summary": "How to size the actual financial opportunity in a readmission reduction initiative before a vendor sizes it for you.",
      "answer": "To size a readmission reduction opportunity, multiply your annual admissions by your current 30-day readmission rate to get baseline readmissions, then multiply by the realistic relative reduction — typically 10 to 20 percent for a well-run program, not the 40 percent in vendor materials — and by the per-readmission financial impact specific to your payer mix and value-based arrangements. Run this before a vendor conversation so you are checking their model against yours rather than adopting theirs.",
      "content_text": "To size a readmission reduction opportunity, multiply your annual admissions by your current 30-day readmission rate to get baseline readmissions, then multiply by the realistic relative reduction — typically 10 to 20 percent for a well-run program, not the 40 percent in vendor materials — and by the per-readmission financial impact specific to your payer mix and value-based arrangements. Run this before a vendor conversation so you are checking their model against yours rather than adopting theirs.",
      "date_published": "2026-07-08T12:00:00Z",
      "date_modified": "2026-07-08T12:00:00Z",
      "authors": [
        {
          "name": "Christian Armstrong, MD, MHSA"
        }
      ],
      "tags": [
        "Operations"
      ],
      "products": [
        "medinsights",
        "qurvia"
      ],
      "reading_minutes": 5
    }
  ]
}