The ability to learn, adapt, and reflect is the foundation of intelligence. Without long-lived temporal awareness and deep context through time, AI is limited.
Reminis is built on extensive research into how the human brain forms, stores, and recalls memories. Those neuroscience insights have been distilled into code — giving AI systems real memory that grows with every conversation.
It remembers over years — not days, weeks, or months.
AI memory today doesn't scale, can't prioritise, and has no concept of time. There's a better way.
A close friend doesn't recite everything they know about you every time you speak. They recall what's relevant right now — and they know when things happened. Reminis gives AI that same ability: the right memories, at the right moment.
Reminis started with cancer. I was diagnosed, and what followed was months of chemotherapy — an endless stream of hospital visits, blood samples, MRIs, CT scans, follow-ups, new medications, changed medications, side effects on top of side effects. Nausea, headaches, mouth problems, fatigue. Every single day brought new details to track, new dates to remember, new places to be.
And all of this during a time of deep anxiety and chemo-induced brain fog — when keeping track of anything felt impossible. I tried every AI assistant available. None of them could keep up. Next session? Gone. Months of context, forgotten.
So I built Reminis. It became the memory I couldn't rely on myself to have. Every day during treatment, I talked to it naturally.
I fought the cancer and beat it. But the real surprise came after — in the doctor's office. My oncologists and nurses were stunned by what the system could do. During follow-up meetings, I could give them instant, precise insight into my treatment history: progress over time, every side effect and exactly when it occurred, which medications I was on when something happened, whether I'd taken a specific pill on a specific day. They could cross-reference my blood samples for any period. They could ask "were you on medication X when this symptom appeared?" and get an exact answer.
All of it from casual, everyday conversations — over a year earlier. No logs, no journals. Just memory.
One vague question. A date-specific, detailed answer drawn from months of conversation.
No manual notes. No tagging. Just natural conversation — and the system pieces it together when it matters.
No journals. No spreadsheets. Just conversation — remembered with perfect detail.
Reminis doesn't just store data. It learns what matters about each user, builds understanding over time, and retrieves the right memories at the right moment.
A user talks to your AI naturally. No special commands, no "save this" — just normal conversation.
Reminis picks up on what matters. A new hobby, a stressful week, a dog's name mentioned once in passing. It decides what's worth keeping.
The way hearing "dad" or "home" instantly triggers a web of connected memories in your brain — Reminis works the same way. Multiple retrieval strategies work together so the right memories surface naturally, not through brute-force search.
Over weeks and months, the AI builds a genuine understanding of who the user is. It gets better the more you talk.
Reminis is the memory engine that makes AI worth talking to twice. It's what turns a stateless language model into something that actually knows you.
Not everything is worth remembering. Reminis distinguishes throwaway comments from meaningful details — the way a person naturally does.
One seamless personality — but the depth of recall adapts to the moment. A quick check-in and a heartfelt conversation draw from different layers of memory, naturally and automatically.
The more conversations happen, the richer the understanding becomes. After weeks and months, the AI knows who you are — not just what you last said.
Every memory is encrypted before it ever touches a database. GDPR-ready architecture. Built for healthcare, finance, and high-sensitivity environments.
Bring your own model — commercial APIs, enterprise accounts, or fully self-hosted open-source LLMs. Reminis works with whatever you choose.
Define how your AI sounds, behaves, and relates to users. The personality layer is fully configurable — from clinical and professional to warm and conversational. Same memory engine, your voice.
Say "I need a new floor mat for the Subie" and Reminis already knows you drive a Subaru Legacy. It searches with that context built in — no need to spell it out. The results are shaped by what it knows about you.
Share an image — a document, a screenshot, a photo — and Reminis doesn't just process it. It remembers what it saw. Ask about it weeks later and it's still there, in full detail.
Reminis is designed to run wherever you need it. Self-host the entire stack, use your own LLM provider, keep the database on your own servers. Your data never has to leave your environment.
Run Reminis on your own servers. The entire system — API, memory engine, database — lives in your infrastructure.
Use enterprise AI accounts that don't train on your data. Or self-host open-source models for complete isolation. Your choice.
Customer data stays in a customer-owned database. Full control over where memories live, how they're backed up, and who has access.
Integrate Reminis into any application with a simple API. Send messages, retrieve conversations, and let the memory system handle the rest.
Authentication, encryption, and multi-tenant isolation are built in.
JWT-based authentication with full multi-tenant support. Each user's memories are isolated and encrypted.
POST a user message. Reminis retrieves relevant memories, composes context, and returns an AI response — all in one call.
Behind the scenes, the system decides what to remember from the conversation. No manual tagging. No extra API calls.
Paginated conversation history and memory management endpoints. Build your own frontend, dashboard, or integration on top.
Interested in integrating Reminis into your product?
Get in TouchAn AI that tracks symptoms, medications, and emotional wellbeing across months of treatment. Patients talk naturally, and the system builds a detailed history their doctor can query later.
Not just your name and your job. Your sister's name, that you're moving next month, that you've been stressed about it, and that you prefer tea in the evening. The details that make a conversation feel human.
Characters that recall choices, conversations, and relationships across sessions. The bartender who remembers your usual. The companion who knows what you've been through together.
Support AI that remembers every previous interaction. The customer's history, their setup, what was already tried. First-contact resolution goes up. Handle time goes down.
AI that remembers what a student has mastered, where they struggle, and how they learn best. It adapts its teaching style over weeks and months — not just within a single session.
An AI companion for elderly residents that knows their name, family, life stories, and daily routine. Every conversation picks up where the last one left off — even weeks later.
An AI that tracks emotional patterns, coping strategies, and progress over months. It notices when someone's been down three weeks in a row — context a weekly session can miss.
A legal assistant that remembers every detail of a case across months of preparation. Witnesses, dates, precedents, strategy notes — recalled instantly when a question comes up in court prep.
Tracks injuries, personal records, diet changes, and training phases over time. Knows you tweaked your shoulder last month and adjusts recommendations without being told twice.
An agent AI that remembers every property shown, every reaction, every deal-breaker mentioned in passing. "She liked the kitchen in the Elm Street place but said the commute was too long" — six months ago.
An AI buddy that remembers a new employee's questions, blockers, and progress through onboarding. Knows they struggled with the deploy pipeline in week 2 and follows up in week 4.
An AI advisor that remembers your goals, risk tolerance, life changes, and past decisions. It knows you're saving for a house, just had a kid, and switched jobs — and adjusts its guidance accordingly.
Not just for humans. M2M integrations can feed context into Reminis — IoT sensors, monitoring systems, automated pipelines — and query accumulated knowledge later. A system that remembers what happened last deployment, last outage, last anomaly.
Companion is our hosted chat app — a live interface where invited users can experience Reminis firsthand. Talk to it today, come back next week, and feel the difference real memory makes.
Companion is currently invite-only. It uses vendors such as OpenAI, meaning conversation data is processed through their servers.
For enterprise or self-hosted deployments, you choose the model provider and infrastructure — including fully private, air-gapped setups.