A photo gives an agent visual evidence. GoPic turns that evidence into grounded place candidates — then exposes the workflow as WebMCP tools.
The human supplies the visual clue. The agent can use GoPic's registered WebMCP tools to inspect the current sign, analyze it, compare candidates, and return a grounded place result.
The UI is intentionally human-friendly. The same workflow is exposed to compatible agents through WebMCP.
Instead of asking an agent to guess where to click, GoPic registers
structured actions directly with document.modelContext.
Read the current human-provided photo context and location hint.
read-only · current page state
Run the GoPic sign-analysis workflow and return structured findings.
vision → OCR → candidate grounding
Inspect candidate places and the evidence used to distinguish namesakes.
read-only · grounded candidates
Text alone can point to the wrong business. GoPic keeps location and candidate evidence in the loop.
Travelers should not need to type characters they cannot read just to find reviews or place information.
The human chooses what matters in the physical world; the agent handles the structured web workflow.
Real registered tools expose a non-trivial visual-to-place workflow.
One coherent web experience for both humans and compatible AI agents.
A concrete travel problem: turning an unfamiliar real-world sign into the correct place.
Bridge the physical world and the open web through a visual agent interface.
GoPic already explored visual sign recognition and place search as a mobile product.
GoPic exposes the sign-to-place workflow directly to web agents through WebMCP tools.