Fog of You

Inspiration

We’re tired of feeling like the internet “knows us” too well. Everything we click becomes a permanent signal that fuels ads, feed manipulation, and creepy personalization. Blocking trackers helps, but a lot of profiling still happens from behavioral patterns. We wanted a privacy tool that doesn’t just hide you, but makes your data footprint less useful by intentionally breaking the profile that gets built about you.

What it does

Fog of You is a local desktop app that runs autonomous “persona agents” that browse harmless content unlike you in the background. Over time, these agents generate realistic decoy activity in separate cookie profiles so trackers and data pipelines see conflicting interests and lower-confidence signals.

• Create 2–5 fake personas (interests, cadence, session length)
• Run “fog sessions” on a schedule or only when idle
• Browse only from a strict safelist (no logins, no forms, no downloads, no ad clicking)
• Show a dashboard with session history, topic diversity, third-party domain contacts, and persona separation

How we built it

• Desktop app UI for persona setup, scheduling, and monitoring
• Playwright-driven Chromium to run automated browsing sessions locally
• Separate persistent browser profiles per persona (isolated cookies, cache, local storage)
• A lightweight “session planner” that generates searches + page visits + dwell/scroll behavior per persona
• Strict safelist + rules engine to prevent risky actions (no logins, no purchases, no sensitive sites)
• Network logging to track third-party domains contacted and quantify how much “noise” we’re generating

Challenges we ran into

• CAPTCHAs / anti-bot systems: many sites detect automation quickly, so we had to tune behavior and choose sources that don’t immediately block automation
• Avoiding logged-in identity graphs: if you’re logged in, fogging can backfire by reinforcing your real profile, so we enforced rules like no logins and isolated persona storage
• Safelist selection: finding domains that are safe, stable, and automation-tolerant took more effort than expected

Accomplishments that we're proud of

• Pivoting from the initial agent-browsing idea into a clear privacy product: behavioral obfuscation that’s local and controllable
• Building true persona separation with persistent isolated profiles so it actually looks like different “people,” not random spam
• Designing a safelist + ruleset that keeps the system defensible (no ad fraud, no sensitive categories, no form submissions)
• Turning an abstract concept (“cloud my data trail”) into something you can see live with metrics

What we learned

Sometimes the best use case isn’t the first one you imagine. The most creative builds aren’t always whimsical — they’re the ones that flip a familiar system on its head. We also learned that privacy products live or die by trust: constraints, transparency, and user control matter as much as the AI.

What's next for Fog of You

• Ship a browser extension version so it’s easier to adopt and feels less like “mysterious background automation”
• Explore a Firefox-based approach (containers / profile isolation) for stronger privacy primitives and cleaner persona separation
• Improve anti-bot resilience ethically by expanding safelists, adding better session realism, and giving users clear controls (idle-only, bandwidth caps, one-click wipe)
• Add a simple privacy-impact view that explains what changed and why it matters

Built With

• css
• gemini
• node.js
• playwright
• react
• typescript