BMRC Hospital: multilingual AI feedback routing
By Anurag Srivastav · AI Automation Engineer
I built an AI feedback workflow for BMRC Hospital that processes multilingual patient feedback, analyzes sentiment, routes tickets and sends Gmail notifications. The documented outcome is automated feedback routing; no accuracy score or time-saving benchmark is published here.
Language in. A structured handoff out.Illustrative system map
The problem
Feedback written in different languages needs interpretation before it can reach the right operational team. The engineering task was to connect language processing to ticket routing and notifications, rather than produce a standalone sentiment score that someone would still have to move between tools.
What I implemented
Built and deployed the multilingual feedback agent during my AI Automation Engineer role at Smartians AI. My portfolio identifies n8n, OpenAI, Supabase and Gmail API as the project stack.
Tools: n8n · OpenAI · Supabase · Gmail API
Workflow at a glance
A simplified view of the stages described in the project account.
Language in. A structured handoff out.Illustrative system map
Receive multilingual feedback
Explore the workflow
Choose a stage to follow the process.
Technical decisions and boundaries
Connect interpretation to a workflow
n8n provides the workflow layer and OpenAI provides the language-processing step. This separates the model’s interpretation of feedback from the subsequent integration work. The portfolio describes sentiment analysis and routing; it does not publish the prompts, category taxonomy or routing thresholds.
Make the outcome actionable
Supabase and Gmail API are part of the documented stack. A routed ticket and a notification are concrete outputs that an operational team can act on. The public account does not expose patient records, database tables or notification recipients.
Keep the scope clear
This case study concerns patient-feedback operations. It does not describe diagnosis, treatment recommendations or a validated clinical decision system. Language coverage and the handling of ambiguous feedback need to be assessed separately from whether the workflow executes successfully.
Results and supporting evidence
Multilingual sentiment analysis and automated feedback routing with Gmail notifications.
What is documented
This is a first-person project account supported by the experience and project descriptions in this portfolio. No public evaluation dataset, per-language accuracy results, ticket-volume log or independent client verification is attached.
How to validate the outcome
A reproducible evaluation would use de-identified feedback labeled by reviewers, report routing accuracy and unresolved cases by language, and compare time to assignment with the previous process. These are proposed validation steps, not measurements claimed for this implementation.
Limitations
- No list of supported languages or accuracy benchmark is published.
- Sentiment classification and correct ticket routing are different outcomes; one should not be treated as proof of the other.
- The workflow account does not establish clinical performance, a compliance certification or a guaranteed response time.
Project questions
What did Anurag build for BMRC Hospital?
Anurag built and deployed a multilingual feedback agent that analyzes patient feedback, routes tickets and sends Gmail notifications using n8n, OpenAI, Supabase and Gmail API.
What results are available for the feedback agent?
The published outcome is multilingual sentiment analysis and automated ticket routing. No numeric accuracy, time-saving figure or per-language evaluation is published with this case study.
Is this a medical diagnosis system?
No. The documented project processes feedback for operational routing. This case study makes no claim about diagnosis or treatment recommendations.
Source material and related reading
These are my own published accounts, not independent third-party verification.