Self-Hosted AI Clinical Support Agent
Budget: $250 – $750 AUD
I am building a point-of-care assistant that runs entirely on the practitioner’s own hardware. When a GP types in a patient’s history, symptoms, lab values or imaging notes, the agent must:
• generate a differential that accurately diagnoses medical conditions,
• recommend an evidence-based treatment plan, and
• cite the latest peer-reviewed information so the doctor can quickly verify each recommendation.
Scope of conditions
The knowledge base has to span common illnesses, chronic diseases and rarer pathologies alike, so the logic that drives the model cannot be trimmed down to a few high-frequency complaints. I expect the same reasoning depth whether the query is seasonal rhinitis, Type 2 diabetes or a paraneoplastic syndrome.
Keeping knowledge current
Updates should be automated and offline-friendly. The agent needs a scheduled routine that pulls new data from major medical journals, government health databases and recent clinical trial results, folds that information into the local store, and re-indexes without me touching a command line.
Install & hosting requirements
• One-click installer for Windows 10+ laptops and a signed .dmg for recent macOS versions.
• No external calls once installed; all inference and storage stay on the device.
• Resource footprint light enough for a mid-range laptop (8 GB RAM) yet scalable if the clinic later migrates to a server.
Deliverables
1. Installation packages (.exe / .dmg) with a guided setup wizard.
2. A lightweight local database of medical literature plus the auto-updater script.
3. UI/UX that lets the doctor enter free-text notes, review suggested diagnoses, see treatment options and click through to the primary sources.
4. Technical documentation covering architecture, update mechanism and a privacy impact statement.
5. Acceptance test suite: at least 10 real-world clinical vignettes showing correct diagnosis ranking and guideline-aligned therapy.
I’m open to the underlying stack—whether you prefer an on-device LLM (e.g., Llama-based) or a hybrid rule-plus-model approach—but the finished system must meet the functional points above and be usable out of the box by a non-technical GP.
I need a fully automated AI clinical decision support agent for general practitioners. The system must be self-hosted and easy to install on any Windows laptop or MacBook with minimal technical knowledge.
Core Requirements:
• Doctors fill a simple Google Form with patient details + photo description.
• New form submissions automatically trigger the AI via n8n.
• The AI (powered by Grok xAI) analyses the information and returns a structured clinical summary.
• The summary is automatically written back into a Google Sheet in a clean, readable format.
• Each doctor runs their own instance on their own laptop using their own Grok API key.
Must-Have Features:
• One-click or simple installer for Windows and macOS (using Docker or similar).
• Automated setup of n8n workflow.
• Secure handling of Google credentials and Grok API key (no hardcoding).
• Pre-configured medical prompt that produces structured output including:
• Patient Summary
• Likely Diagnosis / Differential
• Best Practice Management (Australian guidelines)
• Medication Considerations
• Red Flags
• Recommended Next Steps
• Quick Doctor Checklist
• Clean output saved to Google Sheet.
• Simple instructions/manual for non-technical doctors (70+ years old).
Technical Stack (Preferred):
• n8n (self-hosted)
• Grok xAI API
• Google Forms + Google Sheets
• Docker (for easy installation)
Deliverables:
1. Complete automated installation package for Windows and macOS.
2. Working n8n workflow with Grok integration.
3. Google Form template + connected Google Sheet.
4. Clear installation guide + user manual for doctors.
5. Source files and documentation.
Out of Scope:
• Hosting on a central server
• Mobile app
• Advanced UI/dashboard (Google Sheet is acceptable)
• Medical liability or regulatory compliance
Please include in your quote:
• Fixed price for full delivery
• Estimated timeline
• Whether you have experience with n8n + Docker + API integrations
• generate a differential that accurately diagnoses medical conditions,
• recommend an evidence-based treatment plan, and
• cite the latest peer-reviewed information so the doctor can quickly verify each recommendation.
Scope of conditions
The knowledge base has to span common illnesses, chronic diseases and rarer pathologies alike, so the logic that drives the model cannot be trimmed down to a few high-frequency complaints. I expect the same reasoning depth whether the query is seasonal rhinitis, Type 2 diabetes or a paraneoplastic syndrome.
Keeping knowledge current
Updates should be automated and offline-friendly. The agent needs a scheduled routine that pulls new data from major medical journals, government health databases and recent clinical trial results, folds that information into the local store, and re-indexes without me touching a command line.
Install & hosting requirements
• One-click installer for Windows 10+ laptops and a signed .dmg for recent macOS versions.
• No external calls once installed; all inference and storage stay on the device.
• Resource footprint light enough for a mid-range laptop (8 GB RAM) yet scalable if the clinic later migrates to a server.
Deliverables
1. Installation packages (.exe / .dmg) with a guided setup wizard.
2. A lightweight local database of medical literature plus the auto-updater script.
3. UI/UX that lets the doctor enter free-text notes, review suggested diagnoses, see treatment options and click through to the primary sources.
4. Technical documentation covering architecture, update mechanism and a privacy impact statement.
5. Acceptance test suite: at least 10 real-world clinical vignettes showing correct diagnosis ranking and guideline-aligned therapy.
I’m open to the underlying stack—whether you prefer an on-device LLM (e.g., Llama-based) or a hybrid rule-plus-model approach—but the finished system must meet the functional points above and be usable out of the box by a non-technical GP.
I need a fully automated AI clinical decision support agent for general practitioners. The system must be self-hosted and easy to install on any Windows laptop or MacBook with minimal technical knowledge.
Core Requirements:
• Doctors fill a simple Google Form with patient details + photo description.
• New form submissions automatically trigger the AI via n8n.
• The AI (powered by Grok xAI) analyses the information and returns a structured clinical summary.
• The summary is automatically written back into a Google Sheet in a clean, readable format.
• Each doctor runs their own instance on their own laptop using their own Grok API key.
Must-Have Features:
• One-click or simple installer for Windows and macOS (using Docker or similar).
• Automated setup of n8n workflow.
• Secure handling of Google credentials and Grok API key (no hardcoding).
• Pre-configured medical prompt that produces structured output including:
• Patient Summary
• Likely Diagnosis / Differential
• Best Practice Management (Australian guidelines)
• Medication Considerations
• Red Flags
• Recommended Next Steps
• Quick Doctor Checklist
• Clean output saved to Google Sheet.
• Simple instructions/manual for non-technical doctors (70+ years old).
Technical Stack (Preferred):
• n8n (self-hosted)
• Grok xAI API
• Google Forms + Google Sheets
• Docker (for easy installation)
Deliverables:
1. Complete automated installation package for Windows and macOS.
2. Working n8n workflow with Grok integration.
3. Google Form template + connected Google Sheet.
4. Clear installation guide + user manual for doctors.
5. Source files and documentation.
Out of Scope:
• Hosting on a central server
• Mobile app
• Advanced UI/dashboard (Google Sheet is acceptable)
• Medical liability or regulatory compliance
Please include in your quote:
• Fixed price for full delivery
• Estimated timeline
• Whether you have experience with n8n + Docker + API integrations