Life Orchestrator

Reclaim your time from the chaos of modern life

Comment

Life Orchestrator: Harmony in Every Dimension

Inspiration

The inspiration for Life Orchestrator stems from the modern "productivity paradox." We have more tools than ever to manage our tasks, yet we feel more overwhelmed and disconnected. Traditional calendars treat our lives as a series of disconnected blocks, ignoring the emotional weight of a relationship "overdue" status or the energy required for "deep work." I wanted to build an agent that doesn't just list what you have to do, but reasons about how those things interact—a tool that prioritizes a 15-minute call with a recovering family member as highly as a corporate board meeting.

What it does

Life Orchestrator acts as an autonomous Chief Operating Officer (COO) for your life. It combines a Kinship Ledger (tracking relationship health) with a Life Inventory (managing career and health tasks). Using Gemini 3's advanced reasoning, it proactively "orchestrates" your day—finding optimal windows for flexible tasks around your fixed anchors, ensuring you never miss a critical meeting or a vital connection with a loved one.

How we built it

The application is built using a Local-First Architecture to ensure complete user sovereignty:

  • Frontend: React 19 with TypeScript and Tailwind CSS for a world-class, responsive UI.
  • Logic Engine: Google Gemini 3 handles all real-time orchestration. It utilizes a thinkingBudget to deliberate over complex schedule conflicts and relationship dynamics.
  • State Management: Data persistence is handled purely via browser localStorage. I implemented a custom storage service to monitor and manage the 5MB quota.
  • Integration: Real-time synchronization with Google Calendar via the GAPI/GSI libraries for bi-directional flow.

Challenges we ran into

The primary challenge was Local Storage Constraints. Storing a multi-day chat history with images within a $5 \text{ MB}$ limit is a significant technical hurdle. The storage formula implemented follows: $$ \text{Total Usage} = \text{Text Content} + \sum_{i=1}^{n} (\text{Compressed Image}_i \times 1.33) $$ The $1.33$ multiplier accounts for Base64 encoding overhead. To overcome this, I built a surgical cleanup utility that allows users to delete data by date and implemented a mandatory image compression pipeline.

Accomplishments that we're proud of

We are incredibly proud of the seamless Tool-Calling Orchestration. Seeing the model independently decide to update a relationship status, save a long-term memory, and then re-propose a 24-hour schedule in a single streaming turn is a massive win for agentic UX. We also successfully built a world-class interface that stays under the browser's storage limits without sacrificing the richness of a multimodal experience.

What we learned

Building this project was a masterclass in Agentic UX. I learned that for an AI to be a true "orchestrator," it must be proactive rather than reactive. Users shouldn't have to ask "What should I do now?"; the agent should say, "I've moved your gym session to 4 PM because your 2 PM meeting ran long." I also deepened my understanding of token budgeting and recursive tool calling within the Gemini ecosystem.

What's next for Life Orchestrator

The next phase involves Native Audio Integration using the Gemini Live API. On the technical roadmap, we are planning to migrate from localStorage to IndexedDB to support much larger life histories and multimodal datasets. We are also developing a sophisticated Token Management Engine to surgically manage context windows as user history grows into the millions of tokens, ensuring the agent stays grounded in years of shared memories without performance degradation.

Share this project:

Updates