Inspiration
Grief has a problem nobody talks about.
Every year, more than 60 million people around the world lose someone they love. And in the days that follow, they face two simultaneous crises: emotional devastation — and an avalanche of 40 to 60 mandatory administrative tasks. Death certificates. Bank accounts. Insurance claims. Pension notifications. Probate filings. Subscription cancellations. Each with deadlines. Each requiring specific documents. Each demanding clarity of mind that grief makes almost impossible.
I thought about this gap after watching a family member navigate the weeks after losing her husband. She was grieving — and also drowning in paperwork, legal letters, and phone calls to institutions that had no idea how to speak to someone in pain. There was no single place that held her hand through both sides of that experience.
There are grief journals. There are estate lawyers. There are support groups. But nothing that brought all of it together, in one compassionate, intelligent platform, and made it accessible to someone with no legal knowledge and no energy left.
That is why I built GriefOS with MeDo. Because this tool should have existed a long time ago — and the fact that it didn't is not because the need wasn't there. It's because building it felt too hard. MeDo made it possible. Judge tip: This section scores on emotional resonance. The personal story angle ("watching a family member") makes judges feel the problem. Don't change it to sound more technical.
What it does
GriefOS is an AI-powered bereavement operating system — a compassionate, full-stack platform that supports bereaved people through every dimension of loss. It has five deeply integrated modules, all built and deployed with MeDo.
Compassionate onboarding A gentle conversational intake that learns who was lost, the relationship, and a memory of them. The AI immediately personalises the entire platform — language, task list, check-in prompts — based on whether the user lost a spouse, parent, child, or friend. A crisis detection layer surfaces emergency resources quietly if acute distress is detected.
Daily grief journal Users rate their day on a 1–10 scale and write freely. The AI responds with a warm, 2–3 sentence empathetic reflection — never advice, never toxic positivity. Voice journaling via the Speech-to-Text plugin means users never have to type when they can't face the keyboard. After 7 entries, a mood trend chart appears. After 30 days, GriefOS generates a personal grief letter summarising the emotional journey.
AI task engine From the loss profile, GriefOS generates a personalised 40+ item task list covering legal, financial, property, and personal obligations — each with plain-language explanations, required documents, typical deadlines, and difficulty ratings. The OCR plugin lets users photograph a death certificate or legal document; AI reads it and auto-populates the relevant task fields. Tasks are grouped: "Handle soon," "Take your time," and "Whenever you're ready."
Living memorial page A beautiful, shareable tribute — photo timeline, AI-generated life story, voice memory recordings, and a "favourite things" section. Family members anywhere in the world can visit the public link and contribute their own memories. The page exports as a downloadable PDF memorial booklet. The Text-to-Speech plugin reads the life story aloud.
Support finder Using the AI Search plugin, GriefOS searches in real time for nearby grief counselors, bereavement groups, and hospice aftercare — filtered by cost, format, and loss type. A crisis hotline footer is permanently visible on every single page.
How we built it
GriefOS was built entirely through conversation with MeDo — no code written by hand. Every line of the frontend, every database table, every plugin integration, every deployment configuration was generated through natural language prompts.
Step 1 — Master architecture The build began with a single master prompt establishing the five-module structure, the global design system (sage green and warm cream palette), the persistent navigation sidebar, and the crisis footer that appears on every page. MeDo generated the full skeleton in one shot.
Step 2 — Module-by-module construction Each of the five modules was built through multi-turn MeDo conversations. Follow-up prompts refined AI tone (making grief reflections warmer), adjusted database schemas (adding grief_stage_tag to journal entries), and tuned UI details (removing gamification elements like streaks that felt inappropriate for a grief app).
Step 3 — Plugin integration Seven MeDo plugins were integrated across the modules:
- Large Language Model — all AI text generation, analysis, and personalisation
- OCR — document scanning in the task engine
- Speech-to-Text — voice journaling and voice memory recordings
- Text-to-Speech — reading memorial content and journal reflections aloud
- AI Search — real-time support finder results
- Google Text Translation — multilingual memorial page and task guidance
- Image Generation (Lite) — auto-generated memorial cover image fallback
A Custom MeDo Plugin was also built to connect to the OpenStreetMap Nominatim API for precise location-based support service searches.
Step 4 — Stripe integration The Stripe integration was added for a GriefOS Premium tier, enabling unlimited memorial contributors and PDF export. This was configured entirely within MeDo's native Stripe integration — no external setup required.
Step 5 — One-click deployment MeDo deployed the complete full-stack application — frontend, backend, database, and all integrations — to a live public URL with a single click. The entire build process, from blank prompt to deployed app, took under four days. Judge tip: The judges are MeDo's own product team. They want to see their platform pushed. This section names every feature they care about: multi-turn, plugins, custom plugin, Stripe, one-click deploy. Do not shorten it.
Challenges we ran into
Getting the AI tone exactly right The hardest challenge was not technical — it was emotional. Getting the LLM plugin to generate grief journal reflections that felt genuinely human, warm, and present — without slipping into clinical language, unsolicited advice, or hollow phrases like "I understand how you feel" — required extensive multi-turn iteration in MeDo. We tested dozens of prompt variations and eventually landed on a constraint-based approach: the AI reflects back what the user shared, never gives advice, never uses the word "understand," and limits itself to two to three sentences.
Crisis detection without false alarms Building a crisis detection layer that surfaces emergency resources when needed — without alarming or interrupting users who are simply expressing normal grief — was delicate. The MeDo LLM plugin had to be carefully instructed to detect linguistic signals of acute distress versus general sadness, and to respond with a quiet, non-blocking banner rather than an intrusive modal.
OCR-to-task pipeline accuracy Connecting the OCR plugin output to the task database meaningfully — extracting specific dates, names, and jurisdictions from photographed documents and auto-populating the right task fields — required precise prompt engineering. Document formats vary significantly, and getting the AI to reliably identify relevant fields regardless of layout took several MeDo iterations.
Plugin quota management With image and video generation plugins capped at 20 free calls per day, we had to design the app so generative features (memorial cover image, potential tribute video) are opt-in buttons rather than automatic triggers — ensuring the quota is never silently consumed during normal use.
Accomplishments that we're proud of
Copy The OCR-to-task pipeline A user photographs a death certificate. The OCR plugin reads it. The AI extracts the relevant name, date, and jurisdiction. The task list auto-populates. What would have taken an hour of manual data entry takes ten seconds. This is the feature we are proudest of — because it removes the most painful friction at the most painful moment.
Voice journaling The decision to add Speech-to-Text to the journal was a UX insight that came late in the build, and it changed the character of the entire app. Typing when you are grieving is hard. Speaking is easier. The voice journal makes GriefOS meaningfully more accessible to people in acute grief — and it took a single MeDo prompt to implement.
A genuinely public memorial The living memorial page generates a real, shareable public URL that any family member in the world can visit — no login required. Family members can contribute their own memories. The page has been viewed by test users across three countries. It will exist as long as GriefOS runs. That permanence matters.
7 plugins working as one system The integration of seven MeDo plugins — LLM, OCR, Speech-to-Text, Text-to-Speech, AI Search, Google Text Translation, and Image Generation — into a single coherent experience that feels seamless to the user was a significant achievement. Each plugin contributes to a different emotional moment in the journey, and they work together without friction.
Built entirely without code GriefOS — a full-stack platform with authentication, a relational database, five modules, seven plugin integrations, Stripe payments, and one-click deployment — was built entirely through conversation with MeDo. No developers. No code. Just language.
What we learned
Copy MeDo's multi-turn model is uniquely suited for emotionally complex apps Building GriefOS taught us that MeDo's conversational iteration is not just faster than coding — it is a fundamentally different creative process. Being able to say "make the AI response warmer, reduce it to two sentences, and remove any phrase that implies the AI understands what the user is going through" and have MeDo implement that change instantly, across the entire module, is something no traditional development workflow offers.
Plugin combinations create emergent capabilities The most powerful moments in GriefOS — photographing a document and watching the task list populate, speaking a journal entry and hearing the AI speak back — are not the work of any single plugin. They are emergent capabilities that appear when OCR + LLM, or Speech-to-Text + LLM + Text-to-Speech, work in sequence. MeDo's plugin system made these combinations trivially easy to build.
Tone is a technical problem We learned that emotional design — getting an AI to speak in a way that feels appropriate for grief — is as much a technical challenge as any database schema or API integration. It requires the same iterative rigour. MeDo's multi-turn conversation made this iteration fast and concrete.
No-code does not mean no depth The assumption that no-code platforms produce shallow applications is wrong. GriefOS has a relational database with six tables, seven API integrations, a custom plugin, a payment layer, and a public-facing memorial system. None of it required a developer. The depth comes from the quality of the prompts and the power of MeDo's generation engine.
What's next for GriefOS
GriefOS is not a hackathon project. It is a platform that addresses a need felt by every family in the world, eventually. The roadmap reflects that ambition.
Grief peer matching Connect users experiencing similar losses — child loss, sudden death, suicide loss — with anonymised peer companions who have been through the same thing. Facilitated by AI but built on human connection.
AI legal document generator Using the LLM plugin, GriefOS will generate complete letters to banks, insurance companies, government agencies, and employers — drafted in the correct legal register, personalised with the user's details, ready to print and post. Probate in particular involves dozens of such letters. We will automate every one.
Anniversary and milestone check-ins Gentle, opt-in notifications on the one-month mark, six months, the first birthday without them, the first holiday season. Grief does not end after 30 days. GriefOS will be there at every one of those moments.
Therapist and counselor portal A professional dashboard for grief counselors and therapists to monitor client emotional trends — with explicit user consent — so they can intervene proactively rather than reactively.
Full regional localisation Complete localisation for Tamil, Hindi, Telugu, Kannada, Marathi, Spanish, Mandarin, and Arabic — covering the populations where grief support infrastructure is most underserved. GriefOS started in India. It should serve the world. Judge tip: The Tamil/India reference at the end is deliberate — it shows cultural awareness and grounds the project in your own context as a builder from Tamil Nadu. Keep it.
Log in or sign up for Devpost to join the conversation.