EDITH Core · Working prototype · Currently building
The interface for AI agents
in the real world.
See. Hear. Remember. Delegate. Act. EDITH combines a context-aware wearable with an orchestration layer that routes intent to the right agents and connected services.
Built from India. Designed for everywhere.
AI-GENERATED CONCEPT · NOT CURRENT HARDWARE01 / THE PROBLEM
Users open separate interfaces, reconstruct the situation as a prompt, and manually coordinate the steps between an answer and an outcome.
The user sees, hears or encounters something.
The user manually reconstructs that context as a prompt.
The user moves between apps to turn the answer into an action.
Mobile apps needed the smartphone. AI agents need their own interface.
EDITH connects context, orchestration, and action in one interaction.02 / HOW IT WORKS
A tap or hold captures speech, visual context, or both as one interaction.
EDITH Core combines the request with vision, memory, and relevant context.
An agent selects permissioned tools and structures the steps needed to help.
EDITH answers directly or completes a connected action—with confirmation when it matters.
03 / WORKING USE CASES
Vision, memory, productivity, commerce, and multi-agent orchestration through the same wearable interaction.
04 / WORKING PROTOTYPE
The current prototype combines wearable capture, multimodal reasoning, memory, agent planning, permissioned tools, confirmations, and spoken responses in one system.
Captures speech, visual context, or both through the wearable.
Recalls useful preferences, facts, and recent interactions when relevant.
Selects validated tools and coordinates multi-step requests across domains.
Searches, drafts, schedules, orders, books, or delegates—with confirmation where required.
WORKING IN THE CURRENT PROTOTYPE
05 / EARLY ACCESS
We’re preparing a small early-user group to test EDITH in real routines. Join the list for prototype updates and first access when testing opens.
06 / BUILD LOG
What has moved from question to working system.
Tested whether inexpensive hardware could provide useful real-world context to multimodal AI.
Combined ESP32 camera, microphone, spoken requests, multimodal reasoning, and speech output.
Added persistent preferences, episodic recall, and structured capture for notes, tasks, ideas, and decisions.
Built permission-aware planning across web search, Gmail, Calendar, Swiggy, and Codex.
Working end-to-end wearable context, memory, planning, and action system.
07 / ROADMAP
The next product milestones, without artificial timelines.
Make capture, response, and follow-up conversations feel faster and more continuous.
Create a focused phone layer for context, confirmations, integrations, and personal controls.
Put a small batch into users’ hands and identify the routines that become genuinely habitual.
Move toward a smaller custom form factor while expanding permissioned integrations.
08 / FOUNDER
“AI becomes far more useful when it understands the situation, remembers what matters, and helps carry the intention through.”
A National University of Singapore Computer Science graduate and full scholar, I’ve built products across web, mobile, LLM agents, hardware, and firmware. I’m now building EDITH from India and speaking with early users, researchers, builders, and investors thinking beyond the screen.
09 / GET EARLY ACCESS
Join the early-user list to hear when prototype testing opens.