Implementation Guide: Generate patient discharge summaries, after-visit instructions, and medication guides

Hardware Procurement

Software Procurement

Prerequisites

• Active EHR system with current support contract (athenahealth, eClinicalWorks, NextGen, Epic, Dentrix, Open Dental, Curve Dental, or equivalent). Document the exact EHR product name, version, and whether it supports FHIR R4 APIs.
• Minimum 10 Mbps symmetric internet bandwidth per concurrent provider using ambient AI capture. Run a speed test from each exam room and the nursing station. If bandwidth is below threshold, ISP upgrade must be completed before go-live.
• Chrome (v120+) or Microsoft Edge (v120+) browser installed on all clinical workstations. Freed and Hathr.AI are browser-based platforms.
• Each provider must have an active iPhone (iOS 15+) or Android phone for ambient encounter capture via Freed mobile app. Verify device compatibility before enrollment.
• Enterprise Wi-Fi infrastructure covering 100% of exam rooms, operatories, and provider offices with signal strength of -65 dBm or better. If existing Wi-Fi is consumer-grade or has dead zones, include Meraki AP deployment in project scope.
• Client must designate a HIPAA Privacy Officer and Security Officer (can be the same person) who will co-sign BAAs and approve AI platform configurations.
• Client must provide a list of all providers, their NPI numbers, specialties, and typical encounter types (new patient, follow-up, procedure, etc.) for template customization.
• Active Microsoft 365 tenant with Global Admin access (for SSO configuration and Intune MDM). If client does not have M365, include provisioning in project scope.
• Firewall must allow outbound HTTPS (port 443) to: *.freed.ai, *.hathr.ai, *.openai.com, *.anthropic.com, *.amazonaws.com, and *.microsoft.com. No inbound port openings required.
• Client must execute BAAs with Freed and Hathr.AI before any PHI is processed. MSP should facilitate but the client (covered entity) is the signing party.
• Existing printer infrastructure assessment — at least one laser printer at each discharge/checkout station capable of duplex printing. If not present, include Brother HL-L2395DW in hardware procurement.
• For dental practices specifically: confirm whether practice uses Dentrix, Eaglesoft, Open Dental, Curve Dental, or Denticon. Dental EHRs have limited API integration, so plan for copy-paste workflow as primary integration method.

Installation Steps

Step 1: Site Assessment and Network Validation

Perform a comprehensive site assessment of the practice. Document all exam rooms, operatories, nursing stations, and checkout/discharge areas. Test Wi-Fi signal strength in each location using a Wi-Fi analyzer app. Measure internet bandwidth at multiple points. Inventory all clinical workstations (OS version, browser version, USB port availability). Document the EHR system, version, and current integration capabilities. Identify the discharge workflow: who creates discharge paperwork today, where is it printed, and how is it delivered to patients.

# Wi-Fi signal strength test (macOS)
airport -s
# Wi-Fi signal strength test (Windows PowerShell)
netsh wlan show interfaces

# DNS resolution test for AI vendor endpoints
nslookup app.freed.ai
nslookup app.hathr.ai
ping -c 5 app.freed.ai

Step 2: Network Infrastructure Deployment (If Required)

Install and configure Cisco Meraki MR36 access points to provide enterprise-grade Wi-Fi 6 coverage across the practice. Create a dedicated VLAN for clinical devices (workstations, iPads, provider phones) separate from guest/patient Wi-Fi. Configure WPA3-Enterprise authentication tied to Microsoft Entra ID (Azure AD) for clinical VLAN. Set up the Meraki dashboard for MSP remote monitoring.

Step 3: Workstation Preparation and Browser Configuration

Update all clinical workstations to ensure Chrome or Edge is at version 120 or later. Install the Freed browser extension on each workstation. Configure browser settings to allow microphone access for Freed. Verify USB audio routing for PowerMic 4 devices. Set Chrome/Edge as default browser if not already.

Step 4: Execute Business Associate Agreements (BAAs)

Before any PHI touches an AI platform, ensure BAAs are fully executed between the practice (covered entity) and each AI vendor. The MSP should facilitate this process but the practice's HIPAA Privacy Officer must sign. Obtain and file BAAs from: Freed, Hathr.AI, and Microsoft (for M365/Azure if not already in place). Store executed BAAs in the practice's HIPAA compliance documentation repository.

Step 5: Freed AI Platform Setup and Provider Enrollment

Create the practice's Freed organizational account. Enroll each provider with their individual credentials. Configure SSO via Microsoft Entra ID if the practice has M365 Business Premium. Set up each provider's specialty profile so Freed generates appropriate note structures and patient instructions for their clinical context.

Step 6: Hathr.AI Setup for Custom Medication Guides and Patient Education

Set up Hathr.AI accounts for the practice's designated content creators (typically 1–2 clinicians plus the practice manager). Configure custom prompt templates for medication guides, post-procedure instructions (dental), and patient education materials. This platform supplements Freed by handling content types that require more customization and formulary-specific detail.

Step 7: EHR Integration Configuration

Configure the connection between AI platforms and the practice's EHR system. The integration method depends on the specific EHR in use. For most SMB practices, this will be a copy-paste or browser extension workflow initially, with deeper API integration as a Phase 2 enhancement.

athenahealth Practices

Epic Practices

eClinicalWorks Practices

• Freed browser extension works with eCW web interface
• Verify eCW version supports browser-based access (V12+)

Dental Practices (Dentrix / Eaglesoft / Open Dental)

• These desktop EHRs have limited browser-based interfaces
• Primary workflow: Copy-paste from Freed/Hathr.AI into EHR clinical notes
• For Open Dental: Use Open Dental API (https://opendentalsoft.com/api/) for custom integration
• For Dentrix: Use Dentrix Developer Portal for write-back (requires Henry Schein partnership)

Universal Copy-Paste Workflow (Works with ANY EHR)

Step 8: Template Customization and Content Configuration

Build and configure custom templates for the three primary content types: discharge summaries, after-visit instructions, and medication guides. These templates ensure AI-generated content matches the practice's clinical standards, uses appropriate medical terminology for the specialty, and produces patient-facing materials at the recommended 6th-grade reading level.

Step 9: iPad Deployment for Discharge Station

Configure shared iPads at checkout/discharge stations for clinician review and approval of AI-generated content before it reaches patients. Enroll iPads in Microsoft Intune (or Jamf Now) for MDM management. Configure Shared iPad mode or Single App kiosk mode for the discharge review workflow.

https://portal.manage.microsoft.com
https://app.freed.ai
https://app.hathr.ai

Step 10: Printer Configuration for Patient Document Output

Install and configure dedicated printers at discharge/checkout stations for printing patient-facing discharge summaries, after-visit instructions, and medication guides. Configure print templates with practice branding.

# replace [printer-IP] with the assigned IP address

http://[printer-IP]/

• Set admin password (document in password manager)
• Configure DNS hostname: discharge-printer-01.local
• Enable duplex printing by default
• Set default paper size: Letter (8.5x11)
• Set default quality: 600 DPI

• Windows: Settings > Devices > Printers > Add Printer > [hostname]
• macOS: System Preferences > Printers & Scanners > + > [hostname]

• Brother HL-L2395DW supports AirPrint natively
• Verify iPad can discover printer on same VLAN

Step 11: Pilot Phase — Limited Provider Rollout

Deploy the complete solution to 1–2 volunteer providers for a 5-business-day pilot period. These providers will use Freed for ambient encounter capture, review AI-generated discharge summaries and patient instructions, and test the Hathr.AI medication guide generator. Collect structured feedback daily.

Step 12: Full Practice Rollout and Staff Training

After successful pilot validation, deploy the solution to all remaining providers. Conduct structured training sessions for clinical staff (providers), administrative staff (front desk, checkout), and practice management. Provide printed quick-reference guides and laminated workflow cards for each workstation.

Training Session Plan (90 minutes total)

Session 1: Providers (45 min)

• Overview of Freed: how ambient capture works
• Live demo: complete an encounter with AI documentation
• Hands-on: each provider completes a practice encounter
• Review and edit AI output: what to check before approving
• Hathr.AI demo: generating medication guides
• Q&A and troubleshooting common issues

Session 2: Administrative Staff (30 min)

• Overview: what the AI system does (and does NOT do)
• Discharge station workflow: iPad review → print → hand to patient
• Troubleshooting: printer issues, Wi-Fi connectivity, app login problems
• Escalation: when to call the MSP vs. handle internally

Session 3: Practice Management / HIPAA Officer (15 min)

• BAA documentation review
• Audit log access and review procedures
• Compliance responsibilities: clinician review requirement
• Annual BAA verification process

Deploy Quick-Reference Materials

Step 13: HIPAA Compliance Documentation and Audit Preparation

Complete all HIPAA documentation requirements for the AI implementation. Update the practice's Security Risk Assessment to include the new AI platforms. Configure audit logging and establish the review schedule.

Custom AI Components

Medication Guide Generator

Type: prompt A structured prompt template for Hathr.AI (Claude) that generates patient-friendly medication guides from a list of prescribed medications. The guide includes drug name (brand and generic), purpose, dosing instructions, common side effects, serious warning signs, food/drug interactions, and storage instructions — all written at a 6th-grade reading level. This component handles the medication guide output that Freed does not natively generate with sufficient depth.

Implementation

System Prompt

Configure in Hathr.AI workspace settings:

You are a clinical pharmacist educator creating patient-friendly medication guides. You MUST:
1. Write at a 6th-grade reading level (Flesch-Kincaid score 60-70)
2. Use simple, everyday words instead of medical jargon
3. Include specific, actionable instructions
4. Highlight dangerous warning signs in clear language
5. Never provide dosing that contradicts the prescriber's instructions
6. Always include 'Call your doctor if...' and 'Go to the ER if...' sections
7. Format output for easy printing on standard letter paper

User Prompt Template

# fill in bracketed fields before submitting

Please create a patient medication guide for the following medications:

**Patient Name:** [FIRST NAME ONLY - no last name for privacy]
**Date:** [TODAY'S DATE]
**Prescribing Provider:** [PROVIDER NAME, CREDENTIALS]
**Practice:** [PRACTICE NAME]
**Practice Phone:** [PRACTICE PHONE NUMBER]

**Medications:**
1. [MEDICATION NAME] [DOSE] [ROUTE] [FREQUENCY] - for [INDICATION]
2. [MEDICATION NAME] [DOSE] [ROUTE] [FREQUENCY] - for [INDICATION]
3. [Continue for all medications]

**Special Patient Considerations:** [e.g., elderly, pediatric, pregnant, renal impairment, allergies]

For EACH medication, generate the following sections:

### [MEDICATION BRAND NAME] ([generic name])
**What this medicine is for:** [1-2 sentences in plain language]

**How to take it:**
- [Exact dosing instructions in plain language]
- [Timing: with food, empty stomach, bedtime, etc.]
- [What to do if you miss a dose]

**Common side effects (usually not serious):**
- [List 3-5 most common, with plain language descriptions]
- [When these typically go away]

**Call your doctor if you notice:**
- [List 3-5 concerning but non-emergency symptoms]

**Go to the ER or call 911 if:**
- [List serious/life-threatening reactions]

**Important warnings:**
- [Food interactions]
- [Drug interactions with other listed medications]
- [Activities to avoid: driving, alcohol, sun exposure, etc.]

**Storage:** [Temperature, light, moisture requirements]

---

At the end, include:

## Important Reminders
- Do not stop taking any medication without talking to your doctor first
- Keep all medications out of reach of children
- Bring this guide to your next appointment
- **Questions? Call [PRACTICE PHONE] during office hours**
- **Emergency? Call 911**

Example Usage

Input: Metformin 500mg PO BID - for type 2 diabetes, Lisinopril 10mg PO daily - for high blood pressure Patient considerations: 72-year-old, mild kidney disease

Expected output: Two medication sections with age-appropriate warnings about kidney monitoring, hypoglycemia signs, and ACE inhibitor cough, all in plain language.

Discharge Summary Generator

Type: prompt A structured prompt template for generating comprehensive patient discharge summaries from encounter data. This prompt is used in Hathr.AI when Freed's built-in discharge summary output needs augmentation — for example, when a practice wants a more structured format, dual clinician-patient versions, or summaries that incorporate lab results and imaging findings not captured in the ambient encounter.

Implementation:

System Prompt

You are a medical documentation specialist generating discharge summaries for outpatient medical and dental practices. You create TWO versions of each summary: 1. Clinical version — For the medical record, using standard medical terminology, ICD-10 references, and structured clinical language suitable for provider-to-provider communication. 2. Patient version — For the patient to take home, written at a 6th-grade reading level with plain language explanations.

• Never fabricate clinical findings, lab values, or diagnoses not provided in the input
• If information is missing, note it as '[Not provided - verify with chart]'
• Always include follow-up instructions and return precautions
• Include date, provider name, and practice contact on both versions
• Format for clean printing on letter paper

User Prompt Template

Generate a discharge summary from the following encounter information:

**Visit Date:** [DATE]
**Provider:** [NAME, CREDENTIALS]
**Practice:** [NAME, ADDRESS, PHONE]
**Patient First Name:** [FIRST NAME]
**Visit Type:** [New Patient | Follow-Up | Post-Procedure | Urgent Visit | Annual Wellness]

**Chief Complaint / Reason for Visit:**
[Free text from encounter]

**Diagnoses Addressed:**
1. [Diagnosis - ICD-10 if available]
2. [Diagnosis - ICD-10 if available]

**Key Findings:**
- Vitals: [BP, HR, Temp, SpO2, Weight if relevant]
- Exam findings: [Key positive and negative findings]
- Lab results: [Any results reviewed during visit]
- Imaging: [Any imaging reviewed]

**Treatment Plan:**
- Medications: [New, changed, or continued medications]
- Procedures performed: [If any]
- Referrals: [If any]
- Orders placed: [Labs, imaging, etc.]

**Patient Education Provided:**
[Topics discussed with patient]

Output Format

Clinical Discharge Summary (for medical record)

Structured clinical summary with standard medical terminology. Include: Visit date, Provider, Diagnoses with ICD-10, HPI summary, Pertinent exam findings, Assessment & Plan, Medications reconciled, Follow-up plan, Referrals.

Patient Discharge Summary (for patient take-home)

**[PRACTICE NAME]**
**Your Visit Summary — [DATE]**
**Provider: [NAME]**

Dear [PATIENT FIRST NAME],

Thank you for visiting us today. Here is a summary of your visit:

**Why you came in today:**
[Plain language explanation]

**What we found:**
[Plain language summary of key findings]

**Your diagnoses:**
[Plain language names — avoid medical jargon]

**Your treatment plan:**
[Numbered list of what to do next, in plain language]

**Your medications:**
[Simple table: Medicine Name | How to Take | What It's For]

**Follow-up appointments needed:**
[Specific: who to see, when, and why]

**When to call us ([PHONE]):**
[List of concerning symptoms that warrant a call]

**When to go to the ER or call 911:**
[List of emergency symptoms]

**Questions?** Call us at [PHONE] Monday-Friday [HOURS].

We wish you well!
[PROVIDER NAME, CREDENTIALS]

Dental Post-Procedure Instruction Generator

Type: prompt A prompt template specifically designed for dental practices to generate post-operative and post-procedure care instructions for common dental procedures. Covers extractions, root canals, crowns, implants, deep cleanings, and fillings. Generates patient-friendly instructions with specific timeframes, dietary restrictions, pain management guidance, and warning signs.

Implementation

## System Prompt:

You are a dental patient educator creating post-procedure care instructions. You must:
1. Write at a 6th-grade reading level using simple, clear language
2. Provide specific timeframes (e.g., 'for the next 24 hours' not 'for a while')
3. Include concrete dietary dos and don'ts
4. Provide pain management guidance with specific OTC medication options
5. Clearly describe what is normal healing vs. what requires a call to the dentist
6. Never contradict the dentist's specific instructions provided in the input
7. Format for easy printing with clear section headers

## User Prompt Template:

Generate post-procedure care instructions for:

**Patient First Name:** [FIRST NAME]
**Date of Procedure:** [DATE]
**Dentist:** [NAME, DDS/DMD]
**Practice:** [NAME, PHONE]
**After-Hours Emergency Line:** [PHONE or 'Call 911']

**Procedure Performed:** [Select one or more]
- [ ] Simple extraction (tooth #[NUMBER])
- [ ] Surgical extraction / wisdom tooth removal
- [ ] Root canal therapy (tooth #[NUMBER])
- [ ] Crown preparation (temporary crown placed)
- [ ] Dental implant placement
- [ ] Scaling and root planing (deep cleaning) — quadrant(s): [UL/UR/LL/LR]
- [ ] Composite filling (tooth #[NUMBER])
- [ ] Other: [DESCRIBE]

**Anesthesia Used:** [Local only | Local + nitrous | IV sedation | General]
**Prescriptions Given:** [List medications, doses, and instructions]
**Specific Dentist Instructions:** [Any custom instructions from the dentist]
**Special Considerations:** [e.g., patient on blood thinners, diabetic, pregnant, pediatric]

---

## OUTPUT FORMAT:

**[PRACTICE NAME]**
**Post-Procedure Care Instructions**
**Patient: [FIRST NAME] | Date: [DATE] | Dentist: [NAME]**

### What Was Done Today
[1-2 sentence plain language description of the procedure]

### First 24 Hours — What to Expect
[Normal symptoms: bleeding, swelling, numbness duration, discomfort level]

### Pain Management
[Specific OTC recommendations with doses — e.g., 'Take 400-600mg ibuprofen (Advil/Motrin) every 6 hours with food']
[If prescription given: 'Take [MEDICATION] as directed by Dr. [NAME]']
[What NOT to take if applicable]

### Eating and Drinking
**YES — Safe to eat/drink:**
[Specific food examples appropriate to the procedure]
**NO — Avoid for [TIMEFRAME]:**
[Specific foods/drinks to avoid and why]

### Caring for the Area
[Specific hygiene instructions: brushing, rinsing, salt water rinse recipe if applicable]
[Activity restrictions: no straws, no smoking, no vigorous rinsing — with timeframes]

### Day-by-Day Recovery Timeline
- **Day 1:** [What to expect]
- **Days 2-3:** [What to expect]
- **Days 4-7:** [What to expect]
- **Week 2+:** [When things should be back to normal]

### Call Our Office ([PHONE]) If You Notice:
[Specific warning signs: excessive bleeding definition, fever threshold, increasing pain after day 3, etc.]

### Go to the ER If:
[Emergency symptoms: difficulty breathing, uncontrolled bleeding, severe swelling affecting airway]

### Your Next Appointment
[Follow-up timing and purpose — e.g., 'Return in 2 weeks to have your permanent crown placed']

**Questions? Call us at [PHONE].**
**After hours emergency: [EMERGENCY PHONE]**

Multilingual Patient Education Translator

Type: prompt A prompt template that takes any English patient instruction document and translates it into the patient's preferred language while maintaining medical accuracy, plain language, and cultural sensitivity. Supports Spanish, Mandarin Chinese, Vietnamese, Korean, Arabic, and other languages commonly encountered in US medical/dental practices.

Implementation

# Multilingual Patient Education Translator — Hathr.AI Prompt Template

## System Prompt:

You are a certified medical translator specializing in patient education materials. You must:
1. Translate medical content accurately while maintaining plain language (equivalent reading level in the target language)
2. Use culturally appropriate phrasing — not word-for-word translation
3. Preserve all formatting, headers, bullet points, and structure from the original
4. Keep all proper nouns (medication names, doctor names, practice name, phone numbers) in their original form
5. Include both the English medication name AND any commonly known name in the target language
6. Add cultural context notes where relevant (e.g., dietary instructions that reference culturally specific foods)
7. Output a bilingual document: target language primary with English section headers in parentheses for provider reference
8. IMPORTANT: This is a DRAFT translation. Include a footer note: 'This document was generated by AI translation and should be reviewed by a qualified medical interpreter before clinical use.'

## User Prompt Template:

**Target Language:** [Spanish | Mandarin Chinese (Simplified) | Vietnamese | Korean | Arabic | Tagalog | Russian | Portuguese | Haitian Creole | Other: specify]

**Document to Translate:**
[Paste the complete English patient instruction document here]

**Cultural Notes (optional):**
[Any known patient-specific cultural considerations — e.g., 'Patient follows halal dietary restrictions', 'Patient uses traditional Chinese medicine alongside Western medicine']

---

## OUTPUT FORMAT:

[Full document translated into target language]
[Maintain all original formatting and structure]
[Medication names: [TARGET LANGUAGE common name] ([English brand name] / [generic name])]
[Phone numbers, addresses, and proper nouns remain in English]

---
**⚠️ AVISO / 注意 / CHÚ Ý / 주의 / تنبيه:**
[In target language: 'This document was translated using AI technology. Please ask your healthcare provider or a medical interpreter if you have questions about any of the information in this document.']

[In English: 'AI-TRANSLATED DOCUMENT — Review by qualified medical interpreter recommended before clinical use. Generated on [DATE].']

Freed-to-EHR Copy-Paste Workflow Automation

Type: workflow A documented standard operating procedure (SOP) and optional automation workflow that streamlines the process of moving AI-generated content from Freed into the practice EHR. For practices with browser-based EHRs, this includes keyboard shortcut configurations. For advanced setups, includes a simple AutoHotkey or PowerShell script that monitors clipboard content and formats it for EHR paste compatibility.

Implementation:

Standard Manual Workflow (All Practices)

Step-by-Step Process

Keyboard Shortcut Reference

• Freed: Ctrl+Shift+C (copy current section)
• Switch windows: Alt+Tab
• Paste: Ctrl+V
• Save in EHR: Ctrl+S (most EHRs)

Optional: AutoHotkey Clipboard Formatter (Windows Only)

For practices where EHR paste formatting is problematic (e.g., HTML artifacts from Freed web app), this AutoHotkey script strips formatting and pastes plain text:

# AutoHotkey v2 script for plain-text EHR paste and header injection

; Freed-to-EHR Plain Text Paste
; Install AutoHotkey v2: https://www.autohotkey.com/
; Save as: FreedPaste.ahk
; Run on login via Startup folder

#Requires AutoHotkey v2.0

; Ctrl+Shift+V = Paste as plain text (strips HTML/RTF formatting)
^+v::
{
    ClipSaved := ClipboardAll()
    Clipboard := A_Clipboard  ; Convert to plain text
    Send("^v")
    Sleep(100)
    Clipboard := ClipSaved  ; Restore original clipboard
}

; Ctrl+Alt+1 = Quick paste into EHR note field with standard header
^!1::
{
    CurrentDate := FormatTime(, "MM/dd/yyyy")
    header := "--- AI-Assisted Documentation (Reviewed & Approved) - " . CurrentDate . " ---`n`n"
    ClipSaved := ClipboardAll()
    Clipboard := header . A_Clipboard
    Send("^v")
    Sleep(100)
    Clipboard := ClipSaved
}

Quality Checkpoint

Metrics to Track

• Time from encounter end to completed documentation (target: <3 minutes)
• Number of edits required per AI-generated note (target: <5 per note)
• Provider satisfaction score (target: 8+/10 by week 4)

HIPAA Audit Log Review Automation

Type: workflow A monthly workflow for the MSP to review AI platform audit logs, verify BAA compliance, and generate a compliance report for the practice. This is a key managed service deliverable that justifies the HIPAA compliance management fee.

Implementation

Step 1: Export Audit Logs

Schedule: First Monday of each month (or nearest business day) | Time Required: 30–60 minutes per practice | Responsibility: MSP Compliance Analyst or vCISO

Freed

Hathr.AI

Microsoft 365 (Entra ID)

Step 2: Review Checklist

For each audit log, verify the following:

Step 3: Generate Monthly Compliance Report

Save the following as a Word doc in SharePoint using the template below:

AI Platform HIPAA Compliance Report
Practice: [Name]
Reporting Period: [Month Year]
Prepared By: [MSP Analyst Name]
Date: [Report Date]

Executive Summary:
[1-2 sentences: Compliant/Non-compliant, any findings]

Platform Access Summary:
| Platform  | Active Users | Total Logins | Unique Sessions |
| Freed     | [X]          | [X]          | [X]             |
| Hathr.AI  | [X]          | [X]          | [X]             |

Findings:
1. [Finding or 'No findings this period']

BAA Status:
| Vendor    | BAA Signed | Expiration | Status |
| Freed     | [Date]     | [Date]     | Active |
| Hathr.AI  | [Date]     | [Date]     | Active |
| Microsoft | [Date]     | [Date]     | Active |

Recommendations:
1. [Any recommended actions]

Next Review: [First Monday of next month]

Step 4: Deliver Report

Testing & Validation

• AUDIO CAPTURE TEST: In each exam room, start a Freed ambient recording on the provider's iPhone, conduct a 2-minute simulated conversation at normal speaking volume, and verify that Freed produces a coherent transcript with >95% word accuracy. Test with exam room door open and closed. If accuracy is below 95%, troubleshoot Wi-Fi signal strength and ambient noise levels.
• FREED NOTE GENERATION TEST: Using a standardized test script (non-PHI sample encounter for a patient with hypertension and diabetes follow-up), verify that Freed generates: (1) a complete SOAP note with all four sections populated, (2) patient instructions that mention both conditions, (3) medication list that matches what was discussed. Document the time from recording end to note availability (should be <90 seconds).
• FREED PATIENT INSTRUCTION OUTPUT TEST: After a simulated encounter discussing a new medication start, verify that Freed's patient instruction output includes: (1) the medication name, (2) dosing frequency, (3) key side effects to watch for, and (4) follow-up appointment reminder. Verify the output reads at approximately 6th-grade level by pasting into a readability checker (e.g., readabilityformulas.com).
• HATHR.AI MEDICATION GUIDE TEST: Using the Medication Guide Generator prompt template, input 3 medications (Metformin 500mg BID, Lisinopril 10mg daily, Atorvastatin 20mg at bedtime) for a 65-year-old patient with no special considerations. Verify the output includes all required sections (purpose, dosing, side effects, warnings, interactions, storage) for each medication and is written in plain language without medical jargon.
• HATHR.AI DENTAL POST-OP TEST: Using the Dental Post-Procedure Instruction Generator, input a surgical wisdom tooth extraction with local anesthesia and a prescription for Amoxicillin 500mg TID x 7 days and Ibuprofen 600mg Q6H PRN. Verify output includes: bleeding management (gauze biting protocol), dietary restrictions with specific food examples, day-by-day recovery timeline, and clear emergency warning signs.
• MULTILINGUAL TRANSLATION TEST: Generate a standard after-visit instruction in English, then translate to Spanish using the Multilingual Patient Education Translator prompt. Have a Spanish-speaking staff member (or use a professional interpreter) review the translation for accuracy, cultural appropriateness, and readability. Verify medication names are preserved in English with Spanish explanations.
• EHR COPY-PASTE WORKFLOW TEST: Complete the full workflow from Freed output to EHR documentation: (1) Copy clinical note from Freed, (2) Paste into EHR note field, (3) Copy patient instructions from Freed, (4) Paste into EHR patient instruction field, (5) Save, (6) Print patient instructions from EHR. Verify formatting is preserved and no HTML artifacts appear in the EHR. Time the complete workflow (target: <2 minutes).
• DISCHARGE STATION IPAD TEST: From the shared iPad at the checkout desk, log into Freed web app, pull up a test encounter's patient instructions, verify display is readable and properly formatted, and print to the Brother HL-L2395DW. Verify printout is clean, includes practice branding, and the 'When to call' and 'Emergency' sections are clearly visible.
• PRINTER OUTPUT QUALITY TEST: Print a sample medication guide (3+ medications, 2 pages) on each Brother HL-L2395DW. Verify: duplex printing works correctly, font size is readable (12pt body minimum), headers are clearly distinguishable, and the practice logo and contact information appear in the header. Give the printout to a non-clinical person and ask if they can understand the instructions.
• WI-FI COVERAGE VALIDATION TEST: With Meraki dashboard open, walk through every exam room and operatory with a provider's iPhone running the Freed app. Verify signal strength is -65 dBm or better in every room. Start a test recording in the room farthest from the nearest AP and verify the recording uploads successfully without interruption.
• HIPAA ACCESS CONTROL TEST: Attempt to log into Freed and Hathr.AI from a non-authorized account (e.g., a test account not enrolled in the practice). Verify access is denied. Then verify that MFA is enforced by attempting to log in with correct credentials but without completing the MFA challenge. Verify login fails without MFA completion.
• AUDIT LOG VERIFICATION TEST: Perform 3 test actions in Freed (login, generate a note, copy content) and 3 test actions in Hathr.AI (login, run a prompt, copy output). Wait 15 minutes, then check audit logs in both platforms. Verify all 6 actions appear in the audit logs with correct timestamps, user IDs, and action descriptions.
• END-TO-END CLINICAL WORKFLOW TEST: Conduct a complete simulated patient encounter from check-in to discharge: (1) Provider sees patient with Freed ambient capture, (2) Freed generates SOAP note and patient instructions, (3) Provider reviews and edits in Freed, (4) Provider generates a medication guide in Hathr.AI for new prescriptions, (5) Provider copies all content to EHR, (6) Staff prints patient instructions and medication guide at discharge station, (7) Patient instructions are sent to patient portal. Time the entire process and compare to the practice's pre-AI baseline. Target: 50%+ time reduction in documentation.
• FAILOVER TEST: Disconnect Wi-Fi briefly during an ambient recording to simulate a network interruption. Verify that Freed's mobile app buffers the audio locally and successfully uploads when connectivity is restored. If the app loses the recording, document this as a known limitation and instruct providers to use the PowerMic 4 (wired USB) as a backup capture method.

Client Handoff

Client Handoff Meeting Agenda (90 minutes)

Attendees Required:

• Practice Owner / Managing Partner
• HIPAA Privacy Officer / Security Officer
• Office Manager
• Lead Clinician (pilot provider)
• MSP Project Lead
• MSP Account Manager

Section 1: Solution Overview and Results (15 min)

• Demonstrate the complete workflow end-to-end with a live (non-PHI) example
• Review pilot phase metrics: time saved per encounter, accuracy scores, provider satisfaction
• Show before/after comparison of discharge documentation quality and completeness
• Confirm all success criteria have been met: >90% accuracy, >50% time savings, positive provider feedback

Section 2: Training Recap and Ongoing Education (15 min)

• Confirm all providers have completed Freed training and generated at least 10 encounters independently
• Confirm all designated users have completed Hathr.AI training on medication guide and instruction generation
• Confirm administrative staff can operate discharge station (iPad, printer, copy-paste workflow)
• Provide login credentials sheet (sealed envelope to HIPAA officer) for all platform admin accounts
• Review recorded training video location in SharePoint for future new-hire onboarding
• Discuss training needs for new providers joining the practice (MSP offers onboarding sessions at $500/session)

Section 3: Documentation Package Delivery (15 min)

Hand over the following documents (printed and in SharePoint):

Section 4: HIPAA Compliance Review (15 min)

• Walk through all executed BAAs and their terms
• Review the mandatory clinician review policy (AI content is NEVER sent to patients without provider approval)
• Demonstrate audit log access and explain the monthly review process
• Explain the MSP's monthly compliance report deliverable
• Review incident response plan: what to do if a potential PHI breach is suspected in an AI platform
• Discuss the 2025 HIPAA Security Rule update implications and how the MSP will help adapt

Section 5: Managed Services Agreement Review (15 min)

• Review the ongoing managed services scope: platform monitoring, template updates, compliance management, quarterly training
• Confirm monthly service fee and what it includes
• Review SLA: response times for platform issues (4-hour business hours, next business day for non-urgent)
• Discuss quarterly business review cadence to assess AI platform performance and optimization opportunities
• Review escalation path: Tier 1 (MSP help desk) → Tier 2 (MSP AI specialist) → Tier 3 (vendor support)

Section 6: Q&A and Sign-Off (15 min)

• Open Q&A from all attendees
• Client signs project completion acceptance form
• Schedule first quarterly business review (90 days from go-live)
• Celebrate the launch!

Maintenance

Ongoing Maintenance Responsibilities

Weekly (MSP — 15 min/practice)

• Monitor Freed platform status page (https://status.freed.ai) for outages or degraded service
• Monitor Hathr.AI status for availability issues
• Check Meraki dashboard for Wi-Fi health alerts, client connectivity issues, and firmware update status
• Review any support tickets from the practice related to AI documentation tools

Monthly (MSP — 60 min/practice)

• HIPAA Audit Log Review: Execute the full audit log review workflow (see Custom AI Component: HIPAA Audit Log Review Automation). Export logs from Freed, Hathr.AI, and M365 Entra ID. Review for unauthorized access, unusual activity, and compliance gaps. Generate and deliver the monthly compliance report to the practice HIPAA officer.
• User Account Reconciliation: Verify that terminated employees have been deactivated in Freed and Hathr.AI. Cross-reference with M365 Entra ID disabled accounts list.
• Template Review: Check if any prompt templates in Hathr.AI need updates based on: new medications added to practice formulary, new procedures offered, updated clinical guidelines, or provider feedback.
• Software Update Verification: Confirm Freed and Hathr.AI are on current versions (SaaS — auto-updated, but verify no feature changes that affect workflows).
• Printer Supply Check: Verify toner levels on discharge station printers; order replacements proactively.

Quarterly (MSP — 2-3 hours/practice)

• Quarterly Business Review (QBR): Present metrics to practice leadership: encounters documented with AI, time savings per provider, template usage statistics, compliance report summary, any incidents or near-misses, and recommendations for optimization.
• Template Optimization: Based on 90 days of usage data, refine prompt templates for better output quality. Incorporate provider feedback collected during QBR.
• BAA Expiration Check: Review all vendor BAA expiration dates. Initiate renewal 90 days before expiration.
• Network Performance Review: Analyze Meraki dashboard data for Wi-Fi performance trends. Address any coverage gaps that may have emerged (e.g., new exam room added, furniture rearranged).
• New Feature Assessment: Review Freed and Hathr.AI release notes for new features that could benefit the practice. Present recommendations to practice leadership for potential feature adoption.

Annually (MSP — 4-6 hours/practice)

• HIPAA Security Risk Assessment Update: Conduct a comprehensive review and update of the practice's SRA to reflect any changes in AI platform usage, new vendors, staffing changes, or regulatory updates.
• BAA Verification: Per proposed 2025 HIPAA rules, verify that all business associates have deployed required technical safeguards through written analysis. Document verification in compliance binder.
• Disaster Recovery Test: Verify that if Freed or Hathr.AI experiences an extended outage, the practice can revert to manual documentation workflows. Conduct a 2-hour 'AI-down' drill.
• Staff Re-Training: Conduct annual refresher training for all providers and staff. Cover any new features, workflow changes, and HIPAA updates. Train any new hires who joined since last annual training.
• Vendor Review: Assess whether current AI vendors still represent best value and capability. Evaluate emerging competitors (the AI medical scribe market is evolving rapidly). Present findings and recommendation to practice.

Escalation Path:

• Tier 1 — MSP Help Desk (SLA: 4 business hours): Login issues, browser extension problems, printer issues, basic troubleshooting
• Tier 2 — MSP AI Specialist (SLA: 1 business day): Template refinement, EHR integration issues, workflow optimization, compliance questions
• Tier 3 — Vendor Support (SLA: per vendor SLA): Platform outages, AI accuracy issues, feature requests, API issues
• Emergency — PHI Breach (SLA: immediate): If a potential PHI breach is suspected involving an AI platform, MSP security team engages immediately. Notify practice HIPAA officer within 1 hour. Follow breach notification protocol per BAA terms.

Model Retraining / AI Update Triggers

Freed and Hathr.AI handle model updates on the vendor side (SaaS). However, the MSP should monitor for and respond to:

• Vendor announces major model upgrade (e.g., Freed switches from GPT-4 to GPT-5): Test all templates with new model before full deployment. Watch for output quality changes.
• Hathr.AI updates Claude version: Re-test all custom prompt templates for output consistency.
• Practice adds new specialty or service line: Create new templates and conduct provider training.
• Regulatory change (e.g., new CMS discharge documentation requirements): Update templates to comply with new requirements.

Alternatives

DeepCura for athenahealth Practices

Replace Freed with DeepCura ($99–$129/provider/month) as the primary AI documentation platform. DeepCura offers full bidirectional API integration with athenahealth, automatically reading patient context from athenaOne and writing back AI-generated content into HPI, Physical Exam, Patient Instructions, Assessment & Plan, and 14+ EHR fields — eliminating the copy-paste workflow entirely.

Microsoft Dragon Copilot (DAX) for Epic Practices

Custom Azure OpenAI + FHIR Integration

Build a fully custom solution using Azure OpenAI Service (GPT-5.4) with FHIR R4 APIs to read patient data from the EHR and generate discharge summaries, patient instructions, and medication guides. Deploy as a custom web application hosted in Azure with a HIPAA-compliant architecture.

AWS HealthScribe for API-First Approach

Use Amazon Web Services HealthScribe ($0.10/minute of audio) as the clinical documentation backbone instead of Freed. AWS HealthScribe is a purpose-built HIPAA-eligible API that converts patient-clinician conversations into structured clinical notes and extracts relevant medical terms. Build a lightweight web frontend for providers.

BastionGPT for Dental-Only Practices

Replace both Freed and Hathr.AI with BastionGPT ($30–$60/user/month) as a single HIPAA-compliant AI platform for dental practices. BastionGPT includes pre-built dental prompt templates for clinical note summarization, post-operative instructions, patient communications, and treatment plan explanations.

Suki AI for Multi-Specialty Groups

Replace Freed with Suki AI ($199–$399/provider/month) for mid-size multi-specialty medical groups. Suki offers voice-command-driven AI documentation with the ability to generate notes, patient instructions, and even create orders via voice, across 100+ medical specialties.