-
-
Role-based entry points for startups, investors, incubators, schemes, and the AI Virtual Pitcher
-
Personalized feed of Ministry of AYUSH schemes with deadlines, funding amounts, stream filters, and save/share.
-
National impact metrics above a startup directory with advanced filters, verification badges, and compare/save/share
-
Filter by category, technology, state, and stage. Cards show tags, funding, founding year, and founder.
-
Spotlight carousel of verified startups with stage, funding, location, and quick access to details.
-
Closing call to action: register startups, find investors, join the ecosystem, or try the Virtual Pitcher
-
Interactive India map: pick a state to see registered startups, top innovator, total funding, and incubators.
-
AI recommendations rank startups by match percentage across Category, Trending, Recently Funded, and Investor Picks.
-
Founder success stories with funding raised and real-world impact numbers.
-
Portfolio view with amount invested per startup and performance trend sparklines.
-
Connection request inbox where investors can accept, decline, or reply to founders
-
Shortlist tracker showing each startup’s pipeline status: Shortlisted, Due Diligence, or Contacted
-
Search AYUSH startups by name, tech, founder, or state, and filter by stream: Ayurveda, Yoga, Unani, Siddha, and more.
-
Investor dashboard: set a thesis by sector, ticket size, and stage, then get startups ranked by fit scor
-
Live AI interview: adaptive questions by voice or text, with camera feed and real-time transcript.
-
Landing page: a one-stop hub connecting AYUSH startups, investors, incubators, and government agencies.
Inspiration
India's AYUSH sector (Ayurveda, Yoga & Naturopathy, Unani, Siddha, Homeopathy) is growing fast, but its startups face a broken pipeline. Incubators and ministries receive thousands of unstandardized pitch decks that take weeks to review by hand. Investors hesitate because traditional-medicine ventures rarely map cleanly to standard biotech metrics. A static deck also can't show a founder's domain conviction or regulatory readiness. We wanted a system that evaluates every AYUSH startup fairly, quickly, and consistently, and sends each one to the right next step.
What it does
Startup-AYUSH Portal is an AI-native evaluation and routing platform:
- Deck ingestion: founders upload a PDF/PPT and Gemini extracts market, product, traction, and financial signals.
- Competitor verification: claims are checked against a local FAISS knowledge base of AYUSH ventures, with live web search (Serper/Tavily) as a fallback.
- Live adaptive interview: a Groq-powered AI asks 5 follow-up questions that target weak spots in the deck and in the founder's earlier answers.
- Calibrated scoring: pitch (/50) and founder profile (/40) are normalized into one index out of 30.
- Automatic routing: strong ventures go to the Investor Deal Flow, developing ones go to Incubator Mentorship.
- Ecosystem portals: startup directory, investor dashboard, government schemes feed, and founder workspace.
How we built it
A 4-stage agent pipeline runs on FastAPI (Python):
- Idea Intake Agent: Gemini multimodal ingestion plus FAISS RAG.
- Interview Orchestrator: Groq for low-latency adaptive questions, with transcripts split into idea and profile streams.
- Pitch and Profile Evaluators: rubric-based scoring.
- Scoring and Feedback Agent: normalization, routing, and a report with strengths, vulnerabilities, and 90-day recommendations.
A Node.js/Express gateway handles auth (email OTP with SHA-256 hashing, JWT access tokens, HttpOnly refresh cookies, multi-device logout) and pitch storage in MongoDB. The frontend is five React + Vite + Tailwind apps plus a standalone interview client and a canvas-animated landing page.
The scoring model is proportional:
$$\text{Total} = \left(\frac{\text{Pitch}}{50}\right) \times 19.5 + \left(\frac{\text{Profile}}{40}\right) \times 10.5$$
A total of $18.0$ or higher routes the venture to investors. Anything below goes to mentors.
Challenges we ran into
- Unreliable hackathon networks: we built an automatic in-memory failover so the backend keeps running when MongoDB is unreachable.
- Knowledge base gaps: novel AYUSH ideas missed the FAISS index, so we added a live web search fallback.
- LLM rate limits and failures: we added timeouts and retry logic around Groq and Gemini calls.
- Email delivery: OTP codes fall back to console logging when SMTP isn't configured.
- Coordinating many services: one master script launches the gateway, AI API, and all frontends together.
Accomplishments that we're proud of
- A full pitch-to-routing evaluation that runs end to end in [X seconds, put your measured time].
- An interview that adapts to the deck instead of asking generic questions.
- A transparent, reproducible scoring formula in place of subjective review.
- A demo that keeps working even when the network or database goes down.
What we learned
How to chain multimodal and low-latency LLMs in one pipeline, how RAG and live search complement each other, and how to design for failure from day one.
What's next for Startup-AYUSH Portal
- Expand the knowledge base with verified AYUSH enterprise data.
- Add voice-based interviews.
- Integrate real Ministry of AYUSH scheme eligibility and applications.
- Add reviewer dashboards for incubators and a human-in-the-loop override for scores.
Log in or sign up for Devpost to join the conversation.