I built a hidden AI overlay that's invisible during screen sharing — here's how and why

try { if(localStorage) { let currentUser = localStorage.getItem('current_user'); if (currentUser) { currentUser = JSON.parse(currentUser); if (currentUser.id === 3901592) { document.getElementById('article-show-container').classList.add('current-user-is-article-author'); } } } } catch (e) { console.error(e); } Vishal Kamble Posted on Apr 28 I built a hidden AI overlay that's invisible during screen sharing — here's how and why #showdev #beginners #webdev #productivity The Problem Nobody Talks About There's a moment every IT pro, sysadmin and help desk technician knows intimately. You're screen sharing. Ten people are watching. Something breaks unexpectedly. And suddenly you're on your own — no safety net, no way to quietly Google the error, no AI tool you can reach for without your entire audience watching you do it. Every AI tool in existence assumes you have a quiet moment to switch context. Nobody built one for when you're actually in the fire. So I built it myself. What is GhostPilot? GhostPilot is a Windows desktop overlay that sits on your screen during any screen sharing session — completely invisible to Teams, Zoom, Webex and any other capture tool. You see it. Nobody else on the call does. One hotkey toggle. Full AI chat window. Paste an error, describe the issue, get step by step guidance — all while your audience sees only your desktop, your terminal, your RDP session. Never GhostPilot. The Technical Core The entire idea hinges on one capability — making a window genuinely invisible to screen capture at the OS level. Not minimised. Not alt-tabbed away. Actually invisible while sitting right on your screen. I won't go into the specific implementation details here, but Windows exposes native APIs that allow a window to be excluded from screen capture entirely — at the OS level, not the application level. This means it doesn't matter which screen sharing tool is running. The exclusion happens before the capture ever reaches the application. Once I confirmed that worked, the rest was execution. The Stack Language: C# UI Framework: WPF (.NET 8) AI: Third party AI API with full conversation context Security: Windows DPAPI for local credential encryption Build: Visual Studio 2022 I chose C# + WPF deliberately. This needed to be native Windows — performance, OS API access and reliability mattered more than cross-platform flexibility. WPF gave me full control over the window behavior at a level that Electron or other frameworks simply couldn't match for this use case. What I Learned Building Solo 1. Validate the core technical assumption first Before designing any UI, before thinking about monetisation, before writing a single line of application logic — I proved the invisible window worked. That was the only thing that mattered. Everything else was just execution once that was confirmed. 2. Scope ruthlessly My V1 has exactly what it needs and nothing more. No cloud sync. No user accounts. No mobile. No Mac support. Every feature I didn't build in V1 is a feature I didn't have to debug, test or maintain while trying to ship. 3. The niche is the moat GhostPilot isn't for everyone. It's for IT pros on screen sharing calls. That specificity is a feature not a bug. A focused tool for a specific painful moment beats a general tool that does everything for nobody. 4. Ship before it's perfect The version I launched has rough edges. The version I was planning to launch before I talked myself out of perfecting it would have taken three more…

This excerpt is published under fair use for community discussion. Read the full article at DEV Community.