HaqSe

HaqSe (हक़ से) — India's AI Consumer Rights Champion Pitch: Millions of Indians lose to unfair landlords and bureaucracy daily. HaqSe is an AI chatbot that instantly turns plain-language problems into ready-to-file legal documents and RTIs.

What Inspired Us Seeing grassroots social reform efforts up close made one thing painfully clear: ordinary citizens don't lose legal battles because they are wrong; they lose because the system is intimidating, expensive, and hostile. While the US has tools like DoNotPay, Indian consumers are stuck navigating bureaucratic mazes or portals that suffer from terrible UX and hidden costs.

We realized that civic participation is strained simply because fighting back is too hard. We built HaqSe to be the bridge between the average citizen and their legal rights, turning plain-language frustrations into actionable governance.

How We Built It We optimized for speed and user experience. We built a React-based web app featuring a chat-first, mobile-optimized interface. The architecture is intentionally lean—no heavy databases, just a seamless front-end talking directly to an LLM API.

We focused on four high-impact flows:

RTI Drafter 2. National Consumer Forum Complaints 3. Landlord-Tenant Legal Notices 4. TRAI Telecom Escalations

For the UI, we utilized a card-based category selection that leads into a guided Q&A, terminating in a document output. To reduce the inherent anxiety people feel when dealing with legal disputes, we kept the design clean and minimal, utilizing calming accents like lavender rather than aggressive, traditional "legal" colors.

Challenges We Ran Into Our biggest hurdle was the prompt engineering and contextual boundary-setting.

Preventing Hallucinations: We had to strictly constrain the system prompts with precise Indian legal templates (like specific Sections of the Rent Control Act or proper DISCOM grievance structures) to ensure the AI didn't accidentally output US or UK legal standards.

Flow Control: Keeping the chatbot from turning into an endless conversation. We had to engineer the AI to ask only 3-4 specific, necessary clarifying questions (dates, amounts, order numbers) and then immediately generate the document.

What We Learned This project reinforced a fascinating sociological reality: civic participation isn't just about voting; it's about accessibility. Bad UI and bureaucratic jargon actively disenfranchise people. We learned that by simply fixing the "interface layer" of governance, you can dramatically shift the balance of power back to the individual.

Technically, we learned how incredibly fast you can ship a high-value MVP when you rely entirely on an API-to-UI pipeline, stripping away unnecessary backend complexity to focus purely on solving the user's immediate problem.

Built With

• claude