Robin Assistant

💡 Inspiration My mum, a clinician who works with elderly patients, told me something that stuck: older people living alone will book GP appointments they don't medically need, just to have someone to talk to that day. That's not a healthcare problem. It's loneliness in a healthcare costume.

The more I looked, the clearer the real problem got. Living alone is a full-time admin job — medication, appointments, groceries, birthdays — and carers aren't there 24/7. Worse, the admin of reconnecting with people (booking the bus, remembering the date, making the call) becomes a barrier too high to climb, so people stay home. And the technology meant to help keeps leaving this generation behind.

🕊️ What it does Robin is an empathetic admin assistant that gives older people back the time and energy everyday logistics steal — so they can spend it on the people and things they love.

It calls each morning with a warm, brief check-in, then quietly keeps life running:

Handles the load — medication reminders, appointments, groceries, follow-ups ("how did tea with Robert go?"). Options, not orders — for any task it offers choices (order it, take the bus, ask the carer) and never decides for you. It protects independence instead of taking it. Remembers your week — it understands the person, it doesn't just answer them. That continuity is the difference between a companion and a command line. Keeps loved ones in the loop — family set it up and send reminders through it ("your new shoes are ready"), staying in their person's corner from anywhere. Calibrated to the person — pace, volume, and repetition are tuned to each individual's accessibility needs. Crucially, Robin is not a counsellor. When someone feels low, it doesn't process their feelings — it does the caring thing: it lightens the load and makes reconnecting effortless ("their granddaughter's birthday is Saturday — shall I call her for you now?"). The empathy is in the relief.

🔧 How I built it Voice — ElevenLabs Conversational AI, with Robin's persona and the person's pre-seeded memory in the system prompt. Family dashboard — built fast in Lovable: set up a loved one, send reminders Robin delivers. Brain — structured memory (profile, meds, appointments, contacts, events) plus an event log, and a "morning briefing" that surfaces follow-ups and anything overdue. Deliberately not a knowledge graph — structured records are simpler and more reliable here. LLM — the hackathon's OpenAI-compatible endpoint. 🧗 Challenges I ran into Empathy without counselling. The hardest call: how is an "empathetic agent" empathetic if it isn't a therapist? I landed on empathy-as-relief — care shows up in lightening the load and removing the barriers to human connection, not in emotional processing. Scoping in three hours. I cut hard to one demo that carries the whole thesis, and mocked what I couldn't build. Not over-promising. A voice agent that says "I'll sort that" needs somewhere for that to actually go — so follow-ups route to the family app.

🌟 Accomplishments that I'm proud of A working, two-sided voice demo in the time — and one that feels right. Options-not-orders, memory-driven follow-ups, and the family-in-the-loop moment all landed. Most of all, I think I got the empathy honest: dignity-first, grounded in real clinical insight, and careful about its limits.

📚 What I learned In an admin product, empathy lives in the delivery and the relief — not in a sympathetic script. The right tool usually isn't the fanciest (a structured store beat a knowledge graph; a purpose-built voice API beat building the loop from scratch). Talking to a real domain expert changed the product for the better at every step. 🚀 What's next for Robin Real integrations (calendar, groceries, transport), a live sync between the family app and the agent, a proper calibration onboarding, accessibility profiles, and gentle escalation to family or carers when something's wrong. And, closer to EverSettled's world: helping people keep their affairs in order — peace of mind now, and far less admin left to their families later. Then: trials with real carers and families.

Built With

• elevenlabs
• lovable