The inspiration for Lotti began with a simple question and a powerful moment of human memory.
I was recording videos with my grandmother, going through her photo albums together. She was 92 at the time, and we came across a photograph of her standing in Iceland, clearly in front of a Geysir.
"Grandma, which year was that? Who were you with?" I asked.
She tried to remember. As her memory began to fail her, I could see the frustration building. But then something remarkable happened. She got up, walked to one of her drawers, and pulled out a folder. Inside were handwritten recordings of all her travels—she had traveled six times per year throughout her life.
She scanned the index, flipped through a few pages, and found what she was looking for. The entire lookup and reading took less than a minute. But what happened next was magical: life returned to her eyes. She was happy, smiling, and the memories came flooding back. She began reporting as if she was there again, describing the amazing trip from Hamburg via the Shetland Islands and Faroe Islands to Iceland with her friend Marlies.
The brief paragraph in her journal had unlocked far more information than what was written on the page. It was 1987 she was recalling—nearly 30 years later—with just a little help from her notes.
In that moment, I thought about my own life. I had been to over 40 countries by that point, but how much did I truly remember? How many details had faded? I wanted to remember my experiences the way my grandmother could—with that spark of life and vivid recall.
This inspired me to build what initially was conceived as a travel journal app with audio recording capabilities. The goal was simple: record experiences as they happen, then later—when AI technology matured—be able to summarize and search through these memories just as my grandmother had done with her handwritten notes, but digitally and more powerfully.
As I developed the initial journal, I realized something important about feedback loops. It wasn't practical to record things now and wait decades to see if they were useful. I needed something with a much shorter feedback cycle.
That's when I recognized that memory augmentation and context recovery are valuable on much shorter timescales:
In everyday work, when someone asks, "Do you have a minute?" my gut reaction is always "Yes, sure." But it's never just a minute—it's also the 20 minutes needed to get back into the zone, to recover the mental context of what I was doing.
What if, when returning to a task, I could get a quick summary of where I was? Context for my brain, context for AI—these aren't very different concepts. With the right context, everything becomes easier.
This realization led to implementing a comprehensive task management system with:
- Different statuses (open, groomed, in progress, blocked, done, rejected)
- Time recording
- Audio notes and screenshots
- Image attachments
- Automatic summarization
Now, whenever I return to a task—whether after an interruption, or days or weeks later when something becomes unblocked—I can quickly get back into context. The AI summarizes my notes, my progress, my thinking at the time.
As AI capabilities exploded starting in 2023, I began integrating these new powers into Lotti. Initially, this meant adding AI features to the task management system—summaries, transcriptions, checklists. For a moment, Lotti risked becoming just another AI-enhanced task list.
But that was never the destination—it was merely a stepping stone. The real vision became clear: Lotti should be a context manager for your entire life and work, a system that understands not just what you need to do, but what you've done, what you've learned, and how it all connects. A digital extension of memory that you fully own and control.
The core philosophy emerged:
- Collect data about what you're doing (tasks, travels, thoughts, learnings)
- Maintain full ownership of that data
- Use AI to make that data useful and accessible
- Never be locked into a single provider
The privacy aspect isn't an afterthought—it's fundamental. Most AI services want to collect your data, build profiles, and lock you into their ecosystem. If you stop paying, you lose not just the service but your entire history and context.
Lotti solves this by:
- Storing everything locally on your devices
- Letting you choose AI providers per category
- Supporting local inference for complete privacy
- Using your own API keys for cloud services
- Never storing your data in the cloud
Today, running large language models locally requires powerful hardware. For example, I now use a laptop with 128GB of RAM and 4TB of SSD. But this is temporary. Over the coming years, local AI inference will become feasible for everyone.
Lotti is built for that future: where AI augments our memory and productivity without compromising our privacy, where we can ask "What did I learn in the first half of 2024?" or "What were the key decisions in Project X?" and get intelligent answers from our own data.
My grandmother's handwritten travel journal unlocked decades-old memories with a simple paragraph. Lotti aims to do the same for the digital age—helping us capture, organize, and recall our experiences and work, whether we need that context in 20 minutes or 20 years.
The difference is that now we can search in any language, get intelligent summaries, and maintain perfect recall—all while keeping our memories and thoughts exactly where they belong: with us.
Building Lotti in public. Follow the journey on GitHub and Substack.