AI WORKSPACE
GEO Agent.
Connect brand observations to practical improvements, with evidence from AI answers you can trace.
View source on GitHub ↗Why I built it
Checking whether an AI answer mentions a brand is only the beginning. You also need to verify the facts, record questions and evidence, and turn findings into content-improvement tasks you can track.
What it includes
The workspace includes content assets, a question library, samples and analysis, a task center and a report center. Reports support human review, version storage and finalized PDFs. A client portal provides read-only access to published reports.
Design decisions
Brand mentions and recommendations are recorded separately, as are task delivery and outcome verification. Links are retained between the fact library, questions, samples, reviews, tasks and reports so the reasoning behind a decision can be revisited.
Implementation
React, TypeScript and Vite power the workspace. A Python API and SQLite store data and state; ReportLab generates reports. The public version includes offline safeguards and requires no API key to try.
Current state
All clients, brands, answers, competitors and metrics are fictional simulations. The public version performs no live sampling or paid model calls. Simulated results do not represent real platform performance; approximate API sampling is also different from rankings in a client application. The source includes local setup instructions.
A closer look
