Support Built for Busy Small Firms

Very easy. The installer runs with one double-click and takes about 2 minutes. A setup wizard walks you through everything: choosing your AI backend, connecting to AWS Bedrock (or setting up local AI), and processing your first test case. Most attorneys set it up themselves without any help. If you have an IT provider, they'll find the deployment straightforward — it's a standard Windows installer with a local web interface. Either way, you're up and running the same day.

MedRecords AI runs as a local web application. Once installed on any computer in your office, your paralegal (or anyone on your network) can access it through their web browser by navigating to the app's URL. There is no separate login for each user — the software is protected by a single password you set during setup. For multi-computer setups, contact us about multi-seat licensing.

Yes. At 15 cases/year, you're currently spending 90–120 hours on manual record review (6–8 hours per case). At a $300/hr billing rate, that's $27,000–$36,000 in lost revenue opportunity. MedRecords AI costs $7,997 once and cuts that time by 90%. You'll see ROI after your first 2–3 cases. Plus, the time you get back lets you take on additional cases you'd otherwise turn away.

A full-time paralegal costs $45,000–$65,000/year, takes 3–6 months to fully train, and maxes out at about 20 active cases. MedRecords AI costs $7,997 once (no recurring fees for the software), requires zero training beyond a 2-minute setup, and helps your existing paralegal handle 3–4x more cases. It doesn't replace human judgment — it eliminates the hours of manual reading so your team can focus on strategy and client service.

Download the installer from your purchase confirmation email and run MedRecordsAI-Setup.exe. The installer handles everything automatically — Python, dependencies, and application files. Once installed, double-click the desktop shortcut or launch.bat and follow the built-in setup wizard. The wizard configures your AI backend (AWS Bedrock or local Ollama), initializes the database, and walks you through your first case.

Windows 10 or later, 8 GB RAM minimum (16 GB recommended if running local AI models via Ollama). Python and all dependencies are installed automatically during setup. If you choose local AI (Ollama), it will also be installed and configured automatically. The application runs entirely in your browser at localhost.

After launching MedRecords AI, navigate to Settings → License and paste your license key into the activation field. Your key was sent to the email address used during purchase. The license activates instantly — no internet connection is required for validation.

Each license covers one machine. If you need to run MedRecords AI on multiple computers, contact us at dan.direnfeld@aiproductivity.dev for multi-seat pricing. We offer volume discounts for firms with multiple workstations.

No. As of v2.2.6, MedRecords AI includes an Automated Dependency Manager. Once Ollama is installed and running, the application will automatically detect it and pull the required medical models (Qwen and GLM) in the background. You will see a "Downloading" status in the logs during the first run.

Nothing is lost. MedRecords AI uses a Resilient Background Worker with checkpoint-based auto-resume. If you close the browser, your computer restarts, or the process is interrupted, the pipeline automatically resumes from the last completed stage when the application restarts. Each stage has its own timeout with automatic retry, and a push-folder retry queue ensures completed reports are always delivered even if the output folder was temporarily unavailable. All progress is stored in the local database for maximum durability.

Yes. As of v2.2.7, MedRecords AI uses AES-256 Encryption at Rest for all files stored in the vault. This includes summaries, audit logs, and de-identified data. Encryption keys are generated uniquely for your installation, meaning even if someone gains access to your files, they cannot read the medical records without your specific machine credentials.

When you process a case, the text is indexed into a local Vector Store on your machine. When you ask a question in chat, the system performs a high-speed local search to find only the most relevant snippets. This data is processed locally by Ollama or sent securely to AWS Bedrock (if cloud AI is enabled) only when requested. No data is used to train external models.

MedRecords AI now includes an Automatic Self-Updater. You can check for updates directly from the Admin dashboard. If a new version is available, the app will download the patch and prompt you to restart to apply the latest security fixes and features.

Upload your document via the Files tab and click Summarize. By default, MedRecords AI uses Ledger mode to extract every medical event into a structured billing ledger with dates, providers, codes, and amounts. The pipeline runs through four stages (Ingestion, Privacy, Extract, Verify) with real-time progress tracking. Once complete, you can view the case briefing card, browse extracted items, generate demand packages, and run case reconciliation.

MedRecords AI supports PDF (both native text and scanned documents via OCR), TIFF, PNG, JPG, DOCX, DOC, XLSX, XLS, TXT, CSV, JSON, XML, LOG, HTML, and medical DICOM imaging files (.dcm, .dicom). Scanned PDFs and image files are automatically processed through the built-in OCR engine with medical terminology correction. Maximum upload size is 100 MB per file.

Ledger (default, 5–12 min) is the forensic processing mode that extracts every medical event into a structured billing ledger with dates, providers, CPT/ICD-10 codes, and dollar amounts. Each item is individually verified by an accuracy critique before being written to the database. Fast (1–2 min) provides a quick narrative overview suitable for initial case screening. Auto (2–5 min) offers a balanced narrative analysis with timeline and gap detection. Thorough (5–15 min) performs deep narrative analysis with cross-referencing and comprehensive source linking. Ledger mode is recommended for demand packages, settlement calculations, and trial preparation.

Yes. MedRecords AI integrates with Ollama, allowing you to run open-source AI models locally on your own hardware. Popular options include Llama, Mistral, Qwen, and other models. This enables fully air-gapped processing where no data ever leaves your machine.

Your one-time license fee includes the complete MedRecords AI software, Standard Support (knowledge base, forum, FAQ, and email support from our team), and all features available at the time of purchase. There are no recurring charges for the software itself.

The only optional recurring costs are Priority Support ($1,500/yr) and AWS Bedrock usage if you choose cloud AI processing (typically $1–$2 per case). All cloud processing goes through HIPAA-eligible AWS Bedrock, covered by the AWS Business Associate Agreement. You can avoid all recurring costs entirely by using Ollama for free local AI processing.

Priority Support ($1,500/yr) includes a 24-hour email response SLA, phone support during business hours, a 4-hour critical issue response guarantee, all version upgrades, a dedicated onboarding call, and quarterly check-in calls. It is especially valuable for firms where downtime directly impacts case deadlines. At just $125/month, it's like having an on-call IT team.

Yes. All medical records stay on your machine and are never uploaded to external servers. If you use cloud AI processing via AWS Bedrock, data is transmitted via encrypted TLS 1.3, processed, and immediately discarded — AWS Bedrock is covered by the AWS Business Associate Agreement for HIPAA-eligible workloads. All local data is protected with AES-256 encryption, and a HIPAA-compliant audit trail logs every access and action.

Only if you choose to use AWS Bedrock for cloud-based AI processing. If you use Ollama for local AI, MedRecords AI operates in a fully air-gapped environment with zero internet connectivity required — ideal for high-security and government environments.

Yes. MedRecords AI is designed for HIPAA compliance through on-premise processing (data never leaves your machine), AES-256 encrypted storage, immutable audit logging (SQLite triggers prevent deletion or modification of access log entries), soft-delete data retention (6-year retention with automated purge), automated breach detection (alerts on bulk view/download/export patterns), PHI-sanitized error messages, and role-based access controls. When using AWS Bedrock, a BAA acknowledgment step is built into the setup wizard. Because data stays on your infrastructure, many traditional HIPAA concerns around cloud vendors are eliminated entirely.

The installer automatically detects and installs Python if it's missing. If you still encounter issues, run the application from a command prompt to see error messages. Common issues include port 8801 already being in use by another application — check with netstat -aon | findstr :8801 and close any conflicting processes. MedRecords AI automatically kills zombie processes on startup, but you can also manually clear the port. If Python was just auto-installed, you may need to close and reopen the command window for PATH changes to take effect.

Download the latest version from your account or purchase confirmation, then extract the files over your existing installation directory. Your data, settings, and license key are preserved automatically — only the application code is updated. No reinstallation or reconfiguration is needed.

Ensure there are no extra spaces before or after the key when pasting. If the issue persists, contact support at dan.direnfeld@aiproductivity.dev with your order confirmation and we will resolve it promptly.

MedRecords AI handles large cases with per-stage timeouts and automatic retry. If any stage times out, click Retry — the pipeline resumes from the last completed stage, so no work is lost. On restart, the application also auto-resumes any interrupted pipelines (up to 3 attempts). For very large documents with hundreds of extraction items, the Verify stage may take several minutes as it batches accuracy checks — this is normal and shown in the progress bar. If timeouts persist, try Thorough mode for narrative processing, or increase the combined_timeout value in config.yaml.

A pipeline watchdog runs in the background and automatically detects stuck pipelines. Each processing stage has its own timeout — if any stage exceeds its limit, it is safely terminated and the pipeline is marked as failed with a Retry button. On restart, the application automatically resumes interrupted pipelines from the last completed stage (up to 3 attempts). If the issue persists after retry, run medcli system doctor --fix to diagnose and auto-repair common issues.

MedRecords AI is designed to handle cases of any size, including 500+ page records with dozens of files. Large cases are automatically split into manageable chunks and processed in parallel. If any chunk fails, the rest continue processing — you can retry only the failed portions. The system uses adaptive timeouts that scale with document size, so large cases receive the processing time they need.

Negligence Detection (Pro) analyzes a processed case for standard-of-care deviations. It identifies timing rule violations (e.g., delayed ER imaging), medication contraindications, documentation inconsistencies, and treatment gaps. Each finding includes a severity level, citation to source records, and a recommendation. You can accept or reject individual findings with one click, and your feedback is stored for continuous improvement.

Auto Task Generation (Pro) reads your completed summary and creates a prioritized list of action items — depositions to schedule, records to request, experts to retain, follow-ups needed, and more. Tasks are sorted by five priority levels (Critical through Routine) with suggested assignees (Attorney or Paralegal). You can export the full task list to CSV for import into your practice management system.

Injury Visualization (Pro) generates an SVG body diagram that maps every diagnosed injury to one of 14 anatomical regions (head, cervical/thoracic/lumbar spine, arms, legs, abdomen, knees, etc.). The diagram uses color-coded overlays to highlight injury locations. You can download the SVG for inclusion in demand packages, settlement brochures, or trial exhibits.

Yes. The Deposition Summarizer analyzes uploaded deposition transcripts to extract key testimony, identify contradictions, flag admissions, and cross-reference statements against the medical records. Upload a transcript through the Depositions tab to get a structured summary of the most relevant testimony.

Yes. MedRecords AI offers built-in integrations with Filevine, CASEpeer, and Litify, plus a generic Webhook connector that works with Zapier, Make, n8n, or any custom endpoint. Configure integrations in Settings → Integrations, test the connection, and push summaries directly into your existing case records with one click.

Yes. Download the pre-formatted Excel template from the Contacts page, fill in your contact details (medical providers, insurance companies, etc.), and use the batch import feature to upload them all at once. Imported contacts integrate with the Email Retrieval Bot for automated records request follow-ups.

The Records tab (v2.5.0) is a dedicated records collection management center with three sub-tabs. Contacts lets you manage all medical providers, custodians, and facilities with sortable tables, filters, batch import/export, and one-click follow-ups. Cases groups contacts by case reference with visual progress bars showing collection status. Activity provides a chronological timeline of all email correspondence for full audit visibility. All contact data is encrypted at rest and every access is logged for HIPAA compliance.

When you add a contact in the Records tab, MedRecords AI automatically sends an initial records request email. If no response is received, you can send tiered follow-ups (polite, firm, final) with one click from the contact detail modal. The system tracks follow-up counts, last contact dates, and collection status for each provider. You can also follow up all pending contacts for an entire case at once from the Cases sub-tab.

The AWS Wizard (v2.5.0, found in Settings) walks you through connecting MedRecords AI to AWS Bedrock for cloud-based AI processing. Enter your AWS credentials (ABSK Bearer Token or Access Key pair), and the wizard validates your connection, verifies model access, and securely stores your credentials in the system keyring. No manual configuration files needed.

The Virtual Paralegal (v2.6.0, Virtual Paralegal tab) is an AI agent that executes real commands on your behalf through natural language chat. Tell it what you need — "check pipeline status," "follow up on pending records for case 2024-Smith," "generate a demand package" — and it translates your request into the correct CLI operations. Safe read-only commands execute instantly; write operations (sending emails, modifying settings) always require your approval first.

Watch folders are directories that MedRecords AI monitors continuously. Any new PDF or document dropped into a watch folder is automatically ingested into the processing pipeline — no manual upload required. Push folders receive completed summary reports automatically. You can configure both through the Virtual Paralegal (paste a path and it asks "watch or push?") or via CLI commands like medcli folders add-watch --path "C:\Cases\Incoming".

MedRecords AI includes a built-in System Doctor that runs 10 automated health checks (database integrity, port availability, API connectivity, disk space, and more). Run medcli system doctor from the command line to diagnose issues, or medcli system doctor --fix to auto-repair common problems. For severe issues, medcli system nuclear-reset --confirm performs an emergency reset (kills zombie processes, clears locks, resets state). You can also ask the Virtual Paralegal to "run system doctor" from the Virtual Paralegal chat.

Ledger Mode (v2.10.0, now the default) is a forensic-grade processing pipeline that extracts every medical event into a structured billing ledger. Instead of producing a narrative summary, it creates individual line items with dates, providers, CPT/ICD-10 codes, billed amounts, and source citations. Each extracted item is then verified by an AI accuracy critique. The result is a comprehensive, auditable medical billing ledger that powers demand packages, settlement calculations, and case briefing cards. Ledger mode processes files through four stages: Ingestion, Privacy, Extract, and Verify.

Accuracy Verification (v2.11.0) is an automated quality gate that runs after extraction in Ledger mode. Every extracted item is checked for source grounding (does the original document support this claim?), factual accuracy, and citation validity. Items that fail verification are flagged with reduced confidence scores or marked for review. For documents with many items, verification runs in batches and typically takes 1–6 minutes. The progress bar on the Summarization page shows this as the "Verify" stage.

Case Reconciliation (v2.11.0) is a per-case quality audit that runs automatically after all documents in a case are processed. It performs five deterministic checks — billing consistency, timeline coherence, diagnosis alignment, provider cross-referencing, and citation validity — plus an LLM-powered narrative cross-check. Results appear as a green/amber/red badge on the case detail page with an expandable panel showing any discrepancies found. You can also trigger reconciliation manually from the case detail page.

The Case Briefing Card is a one-page summary that appears at the top of each case detail view. It aggregates data from all processed documents in the case to show total billed amounts, date range of treatment, primary injuries, treating providers, and key findings. The briefing card is generated automatically from the master case aggregator and updates as new documents are processed. It provides attorneys with an instant snapshot of case value without reading through individual summaries.

When you generate a demand package from a case, MedRecords AI pulls the master billing ledger to create a Master Ledger Summary table directly in the DOCX output. This includes every provider, every visit, billed amounts, and ICD-10 codes — all sourced from the verified extraction data. The demand package also incorporates the case briefing card data for the demand summary section, ensuring the dollar figures in your demand letter match the verified medical billing records.

The audit log is protected by immutable database triggers (v2.10.0) that prevent any DELETE or UPDATE operations on audit entries. Even direct database access cannot modify or remove logged events. Additionally, audit entries include HMAC signatures that detect any external tampering. This ensures a complete, unalterable chain of custody for every access, export, and modification to medical records — critical for HIPAA compliance and litigation defensibility.