AIToday
Large Language ModelsAI in HealthcareAI Business & IndustryAmazon AI BlogPublished: Aug 8, 2026, 04:01 JST

Cohere Health digitizes clinical policies using AWS Bedrock agents

Cohere Health digitizes clinical policies using AWS Bedrock agents

3 Key Points

  1. What happened

    Cohere Health built Cohere Policy Studio using Amazon Bedrock AgentCore to transform static clinical policies into structured, machine-readable data. The application uses a multi-tenant agentic architecture with AgentCore Runtime, Gateway, and Memory to automate prior authorization workflows while maintaining human oversight.

  2. Why it matters

    Prior authorization is one of healthcare's most manual processes, blocking coverage decisions for hundreds of millions of patients annually. By digitizing policies into computable formats, health plans can meet Centers for Medicare & Medicaid Services (CMS) requirements for API-based electronic prior authorization by January 2027 and pursue the America's Health Insurance Plans (AHIP) commitment to achieve 80 percent real-time approvals for electronic prior authorization submissions.

  3. What to watch

    Cohere Health's approach uses modular skills authored by clinical policy experts, evaluated against ground truth datasets for accuracy, completeness, and consistency, then tracked in production via Arize AI. The team monitors for skill degradation over time and prioritizes optimization based on clinical policy analysts' annotations of sample outputs.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

Prior authorization represents a critical bottleneck in healthcare operations because the policies governing it exist in static, unstructured formats—PDFs, documents, and manual records—that resist automation. The challenge is not that the medical reasoning behind prior authorization is flawed; rather, policies vary by clinical area, geography, line of business, and health plan, and they evolve constantly as medicine and technology advance. Historically, health plans lacked systematic tools to manage, analyze, and optimize these policies at scale. By digitizing policies into machine-readable, structured data using standard terminologies, Cohere Health removes this operational bottleneck and enables consistent, computable workflows.

The urgency is regulatory and competitive. The Centers for Medicare & Medicaid Services (CMS) has mandated API-based electronic prior authorization by January 2027, and the America's Health Insurance Plans (AHIP) has set a commitment for 80 percent real-time approvals. Meeting these targets requires automation, but automation of clinical workflows demands domain expertise, transparency, and accountability. Cohere Health's approach addresses this by building a modular skills framework where clinical policy experts author and refine domain-specific capabilities directly. Skills are evaluated against ground truth datasets for accuracy, completeness, and consistency before production deployment, then monitored via Arize AI with feedback from clinical analysts. This design preserves human oversight while enabling rapid deployment across multiple health plans—a multi-tenant architecture that isolates each customer's data while sharing a common runtime infrastructure.

FAQ
What regulatory deadline is driving this project?
The Centers for Medicare & Medicaid Services (CMS) requires health plans to support API-based electronic prior authorization by January 2027.
What approval target are health plans aiming for?
America's Health Insurance Plans (AHIP) commitments require health plans to achieve 80 percent real-time approvals for electronic prior authorization submissions.
How does the system ensure clinical oversight?
Cohere Health uses modular skills authored by clinical policy experts, evaluated against ground truth datasets, and monitored in production with Arize AI. Clinical policy analysts annotate sample outputs to catch errors the automated metrics miss, and the system supports a human-in-the-loop feedback process where teams refine outputs within a governed workflow.
Amazon AI BlogRead Original Article

Get the latest Large Language Models news every morning

For example, today's edition would include:

  • Instinct raises $1B at $10B valuation for personal AI agentSiliconANGLE AI · 2h ago
  • Okta's Wylie: agent security needs shared safeguardsSiliconANGLE AI · 2h ago
  • CoreWeave's top three customers drive 70% of revenue, Vellante saysSiliconANGLE AI · 2h ago

AI-summarized, only the topics you pick: one digest a day via Email, LINE, or Slack.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleUS stocks rise on weak jobs data; Doximity surges 37% on AI search tool