Over the past few weeks, one use case has come up repeatedly in user feedback: locating the exact site of a real-time incident from incomplete online reporting.
In practice, that often looks like questions such as: "Where exactly did the shooting on a beach in Australia take place on December 14, 2025?" Users are not just asking for general context. They need the most precise location possible, fast enough to support active reporting, verification, and investigation.
This has also been evident in our own work. When we publish location breakdowns of real-time news events on social media, the response is enormous. Millions of views later, it became clear that this is not a side use case. It is a serious professional workflow for journalists, investigators, researchers, and open-source intelligence practitioners.
That is why we built GeoSeer Event Mode.
Why we built Event Mode
Before Event Mode, users often approached incident lookup by finding a clear image from a news outlet and treating the problem like a standard image geolocation task. Because GeoSeer already combines web search with geolocation reasoning, that approach often worked.
But it had obvious limits.
- It depended on users having access to a strong visual input.
- It was less effective when reporting was fragmented or early.
- It forced a niche investigative workflow through a system optimized primarily for image-based geolocation.
For real-time incidents, the strongest evidence is often not a single image. It is a scattered trail of partial clues: local reporting, short witness posts, place-name mentions, timeline details, video captions, or a small update from someone on the ground.
How Event Mode works
GeoSeer Event Mode is an agentic web search workflow designed specifically for this problem.
Instead of treating the task as ordinary search, Event Mode performs a deeper investigative process:
- It searches broadly across the web to surface both major coverage and small but meaningful clues, including local reports and social posts.
- It runs multi-step reasoning over those clues to narrow down the most likely location as precisely as possible.
- It cross-references sources against one another to check for consistency and reduce hallucinated conclusions.
- It applies fact-checking throughout the workflow so the final answer is grounded in coherent reporting rather than a single fragile source.
This depth of search and verification is what makes the mode useful. In internal testing, our research team found that this workflow is significantly better suited to real-time event localization than general-purpose AI websites, especially when precision and source quality matter.
Who it is for
Event Mode is built for people who need reliable answers under time pressure:
- Journalists and reporters verifying where an incident actually happened before publishing.
- Investigators and analysts tracing the exact site of a developing event.
- Researchers and OSINT practitioners who need a faster way to assemble and validate fragmented evidence.
What normally takes 30 minutes or more of manual searching can often be compressed into 1 to 2 minutes, with explicit reasoning and source cross-checking built into the workflow.
How to use it
On the web, go to geoseeer.com and select Event Mode in the input bar. Unlike standard image-based workflows, this mode does not require an image. For best results, describe the event as clearly and specifically as possible in text.
For API usage:
- Documentation: https://geoseeer.com/api-docs
- Repo + examples: https://github.com/geoseeer/GeoSeer
Use promo code EVENT for 10 free web search credits.