First Test: Pulling Live Regulatory Data

The first assistant went live on March 25, 2026. The task: check the Federal Register for new HHS publications relevant to hospital finance. No manual trigger. No prompt typed into a chat window. The assistant ran on a schedule, pulled live data from a public API, and produced a summary.

Here is what happened.

What was tested

A regulatory monitor assistant configured to query the Federal Register API (federalregister.gov/api/v1) for recent documents published by the Department of Health and Human Services. The assistant filters results for healthcare and hospital finance relevance, summarizes each item in plain language, and assigns a relevance score from 1 to 5.

The goal: replace the manual process of checking government publication feeds. Hospital finance teams that track regulatory changes currently rely on someone remembering to look, or on expensive subscription services that bundle this with features nobody asked for.

What it found

Five documents published on March 24, 2026. All real. All from HHS.

1. "Submission for OMB Review; 30-Day Comment Request; National Institutes of Health (NIH) Loan Repayment Programs" Document 2026-05738. A notice requesting public comment on NIH loan repayment program data collection. The 30-day comment window means organizations that participate in NIH-affiliated loan repayment have a limited window to respond. Relevance to hospital finance: 2 out of 5. Indirect — affects workforce retention programs at academic medical centers.

2. "Request for Information (RFI): Inviting Comments and Suggestions on a Framework for the NIH-Wide Strategic Plan for Fiscal Years 2027-2031" Document 2026-05734. NIH is soliciting public input on its strategic direction for the next five years. This shapes federal research funding priorities, which flow downstream to hospital research programs and grant-funded positions. Relevance to hospital finance: 2 out of 5. Long-term strategic signal, not immediate operational impact.

3. "Listing of Color Additive Exempt From Certification; Spirulina Extract; Delay of Effective Date" Document 2026-05733. An FDA rule delaying the effective date for spirulina extract color additive regulations following public objections. Relevance to hospital finance: 1 out of 5. Food safety regulation with no hospital finance impact.

4. "Listing of Color Additive Exempt From Certification; Beetroot Red; Delay of Effective Date" Document 2026-05732. Same category as above — FDA delaying a color additive rule for beetroot red. Relevance to hospital finance: 1 out of 5. No operational relevance.

5. "Solicitation of Nominations for Appointment to the Mine Safety and Health Research Advisory Committee" Document 2026-05724. CDC seeking nominations for 10 expert positions on a mine safety advisory committee. Relevance to hospital finance: 1 out of 5. Occupational health adjacent, but no direct hospital finance connection.

The honest result

Two items rated a 2. Three rated a 1. No high-relevance findings on this particular day. That is a normal day. Most Federal Register publications from HHS are not directly relevant to hospital CFOs. The value of the assistant is not that every run surfaces something critical — it is that no publication gets missed on the days when something critical does appear.

A CMS proposed rule on Medicare payment adjustments would score a 5. A new HIPAA enforcement guidance would score a 4. Those documents appear in the same feed. The assistant checks every six hours. The alternative is someone on the finance team remembering to check manually, or not checking at all until a consultant mentions it three weeks later.

What it cost

The Federal Register API is public and free. No authentication required. The data pull cost $0.00.

The AI summary and relevance scoring — processing five documents through the LLM for plain-language summaries and relevance ratings — cost an estimated $0.05. That is the cost of a single run. Four runs per day at six-hour intervals would be approximately $0.20 per day, or about $6.00 per month.

Every cent is tracked. The platform records the exact cost of each LLM call, which model processed it, and when. There is no estimated invoice at the end of the month. The cost is known at the moment it occurs.

How it runs

The assistant operates on a fixed schedule — every six hours, automatically. No one needs to open a dashboard and click a button. No one needs to remember.

The sequence:

1. Query the Federal Register API for recent HHS documents
2. Filter for healthcare and hospital finance keywords
3. Pass matching documents to the LLM for plain-language summarization
4. Score each document for relevance to hospital finance operations (1-5 scale)
5. Store the results with full cost and interaction records

If a high-relevance item appears — a CMS payment rule, a new reporting requirement, an enforcement action — the assistant flags it. The record of what was checked, when, and what it cost exists whether or not anything important was found.

What is next

This was the first assistant. The regulatory monitor covers one data source — the Federal Register. The same pattern applies to:

• CMS Newsroom: Medicare payment updates, coverage decisions, enrollment deadlines
• Board prep assistant: Pulling financial benchmarks and regulatory context before board meetings
• Competitor monitoring: Tracking published financial data from peer hospitals

Each assistant builds on the same platform. Each one has cost tracking, interaction logging, and governance controls from the start. Not bolted on after the fact.

Platform note

This test ran through the Curate-Me platform. The data pull is free — public API, no key required. The AI summary is cost-tracked to the penny. Every interaction has a record: what was sent, what model processed it, what it cost, when it happened. That audit trail exists whether the result was interesting or not.

The regulatory monitor is now running. Six-hour intervals. Zero manual effort. Full cost visibility. The first of several.