Presents
Front-end Interface
How it works
Why choose us
Medical History Session Voice Agent
Patient Dashboard
Doctor Dashboard
SOAP Report with red flags highlighted on Doctor End
System Architecture Design

Sehat Guftagu

AI-powered clinical conversations that help doctors catch what matters early

Inspiration

Many people suffer from serious health complications not because doctors lack expertise, but because critical details are missed early in the process.

This usually happens due to:

Overburdened doctors with only 2–3 minutes per consultation.
Helpline workers and clinics are failing to ask the right follow-up questions.
Patients struggling to explain symptoms clearly, often due to language barriers.

The gap we observed

On average, there is only 1 doctor per 1,000 people globally.
Around 50–70% of medical errors are linked to communication failures.
Language differences (English vs Urdu, Punjabi, Pashto, Sindhi, etc.) further widen the gap.

We imagined a system where patients could calmly explain their problems in their own language, while doctors receive a clear, structured summary before meeting them.

1. Description of the Idea

Sehat Guftagu is an AI-powered health assistant that conducts structured medical interviews with patients in their preferred language and converts those conversations into clear, doctor-ready clinical reports.

Our solution focuses on early symptom clarity, better communication, and time efficiency for healthcare providers. By collecting accurate information before a consultation, we help reduce missed red flags and improve medical decision-making.

2. Target Group

We designed Sehat Guftagu for:

Patients, especially those with language or communication barriers
Doctors and clinicians who face time pressure
Clinics and hospitals managing high patient volumes
Telemedicine platforms needing structured intake data

3. Features & Functions

Core features include:

5–15-minute guided clinical interviews in the patient’s language
Voice and text interaction support
Automated detection of medical red flags
Generation of structured SOAP reports (Subjective, Objective, Assessment, Plan)
Doctor-ready summaries available before consultation

Technology used

AI and Large Language Models for reasoning and translation
Speech-to-Text and Text-to-Speech for voice interaction
Multi-agent orchestration for interview flow and documentation

4. Value Proposition & USP

What makes Sehat Guftagu different:

We focus on pre-consultation clarity, not diagnosis replacement
Patients speak naturally, in their own language
Doctors receive structured, clinically useful summaries
Red flags are highlighted early instead of buried in conversation

Unlike generic chatbots, we are not trying to “act like a doctor”.
We help doctors do their job better by giving them better information upfront.

5. Visualization

Our MVP includes:

A clean, responsive web interface
Real-time progress indicators during interviews
Dynamic sections showing interview stages

6. User Feedback

We validated the idea informally with:

Medical students and junior doctors
General users with prior telemedicine experience

Early feedback

Doctors appreciated receiving structured summaries instead of long patient explanations
Users felt more comfortable speaking in their native language
Many mentioned they remembered details they usually forget during real consultations

This feedback directly shaped our interview flow and report format.

7. Business Model

Possible monetization paths:

B2B licensing for clinics and hospitals
Subscription plans for telemedicine platforms
Freemium model for individuals with paid advanced reports
Future integration with insurance or employer health programs

Our goal is affordability without sacrificing quality.

8. Implementation & Feasibility

Sehat Guftagu is already implemented as an MVP.

Current state

Fully working frontend and backend
Multilingual AI interviews
Report generation pipeline
Deployed using free-tier services

Next steps

Clinical validation with doctors
Improving medical accuracy with fine-tuning
Adding condition-based doctor recommendations
Scaling infrastructure for production use

9. Data Requirements & Privacy

We handle sensitive health data carefully.

Data involved

Patient symptoms (text and voice)
Conversation transcripts
Generated clinical summaries

Privacy & security

No unnecessary data collection
Encrypted storage using PostgreSQL on Supabase
Designed with GDPR/DSGVO principles in mind
Clear consent before any data processing

We aim to remain compliant and transparent as the system scales.

How We Built It

Frontend: Next.js with TailwindCSS
Backend: Next.js API routes with Prisma ORM
AI & LLM: Groq LLaMA models for reasoning and translation
Voice Services: ElevenLabs, Uplift AI, Groq Whisper
Database: PostgreSQL (Supabase)
Agent Orchestration: LangGraph for multi-agent workflows

Challenges We Ran Into

Medical-grade translation between Urdu and English
Balancing AI automation with human oversight
API rate limits on free tiers
Low-latency voice processing

Accomplishments We’re Proud Of

End-to-end multi-agent interview pipeline
Real-time adaptive UI
Multilingual support from day one
Scalable system design

What We Learned

Clear communication saves lives
Healthcare AI must support humans, not replace them
Simple UX matters more than complex features
Free-tier constraints force better engineering decisions

What’s Next

Condition-based doctor recommendations
City-level hospital suggestions
Budget-friendly care options
WhatsApp integration
Secure lifelong medical history tracking

Team Information

Kaleemullah Younas

Role: Full-Stack AI Engineer
GitHub

Muhammad Umer

Role: Web Developer & DevOps
GitHub

Contact

Primary Contact Email:
Contact us

Disclaimer

This is an MVP deployed on free-tier services.
The live version may occasionally face rate limits (HTTP 429).
We plan to upgrade the infrastructure for reliability and scale.