Quick Comparison
| Category | faFAANG | LockedIn AI |
|---|---|---|
| Type | Native Electron desktop app | Desktop app + browser extension |
| Stealth mechanism | WDA_EXCLUDEFROMCAPTURE (named Win32 API) | Unspecified — no API named in docs |
| Stealth workarounds needed | None — works at OS level from launch | Users advised to rename app .exe |
| Click-through overlay | Yes — Coding Mode default | Hotkeys available; click behavior varies |
| Global hotkeys | 21 fully remappable | Some hotkeys, not fully remappable |
| Transcription | Local (Moonshine, on-device) | Cloud-based (hybrid local + server) |
| AI model | Your own ChatGPT account | Their backend (you pay for their compute) |
| Pricing | $49.99 lifetime | $54.99/month |
Area 1: Stealth — The Claim vs The Verifiable API
LockedIn AI's desktop app markets a “True Stealth Mode” that claims no taskbar presence, no Alt+Tab entry, and complete invisibility during screen sharing. These are legitimate claims for a desktop application — in principle.
The problem is that LockedIn AI's documentation never names the mechanism. There is no Win32 API citation. No reference to SetWindowDisplayAffinity. No disclosure of how the Windows Display Infrastructure is instructed to exclude their windows. The stealth is described in marketing terms, not engineering terms.
And tellingly, LockedIn AI's own support documentation publishes a guide titled “Can You Rename LockedIn AI's Desktop App Safely?” — advising users to rename the application's executable so the process name doesn't appear recognizably in Task Manager during screen sharing. This is a user-side workaround. It is not an OS-level guarantee.
faFAANG's approach: faFAANG calls SetWindowDisplayAffinity(hwnd, WDA_EXCLUDEFROMCAPTURE) on every window it creates — control, pane, and pane_history. This is a documented Win32 API in the Microsoft SDK. It instructs the Windows Display Infrastructure compositor to exclude these windows from all capture output at the kernel level. No user configuration. No process renaming. No workaround. The exclusion is applied at window creation and holds for the lifetime of the process.
One calls a verifiable OS API. The other recommends renaming a file. The difference in stealth reliability is the difference between a guarantee and a workaround.
Area 2: Keyboard Control
LockedIn AI does offer some hotkey support for discreet operation. Their documentation lists hotkeys for starting/stopping recording and triggering AI assistance. This is a meaningful feature.
However, the scope and depth differ significantly. faFAANG registers 21 global hotkeys at the OS level via the Windows RegisterHotKey API. These cover every action needed in a live interview:
- ■
Ctrl+S— Start/stop transcription and submit - ■
Ctrl+D— Screenshot the problem statement - ■
Ctrl+M— Switch between Experience and Coding modes - ■
Alt+H— Instantly hide the overlay - ■
Ctrl+Up/Down/Left/Right— Reposition overlay without mouse - ■
Ctrl+Shift+Left/Right/L— Navigate response history
All 21 are fully remappable in faFAANG's settings. If any default key conflicts with your IDE, change it. LockedIn AI's hotkey set is not documented as fully remappable. In Coding Mode, faFAANG's pane window defaults to click-through — non-focusable, passing all input through to CoderPad. Your cursor stays in the editor for the entire interview.
Area 3: Data Handling — Where Your Interview Content Goes
LockedIn AI's AI responses are generated through their backend infrastructure. Their Copilot + Coach system uses cloud AI processing. Your interview audio, transcripts, and prompts travel through their servers to reach the AI model and return to you. LockedIn AI states they don't sell your data, but the data path still runs through their infrastructure.
faFAANG's approach: Speech transcription runs locally via a bundled Moonshine model — audio frames are processed entirely on-device. When a response is generated, it routes through your own ChatGPT/Codex account via the CodexAppServerClient using JSON-RPC. faFAANG is not in the AI response path. Your interview content goes from your machine to your ChatGPT account, same as using chat.openai.com directly.
This also means faFAANG's AI quality is tied to your own ChatGPT subscription. If you're on ChatGPT Plus, you get the latest models on day one. LockedIn AI's AI quality is tied to what they deploy on their backend.
The Cost Comparison
LockedIn AI's Unlimited plans start at $54.99/month. Over a 6-month job search, that's over $329.
faFAANG's lifetime plan is $49.99, paid once. Every feature — WDA_EXCLUDEFROMCAPTURE stealth, local Moonshine transcription, 21 remappable hotkeys, both modes, screenshot grounding — included on every plan including the free trial.
| Duration | LockedIn AI ($54.99/mo) | faFAANG (Lifetime) |
|---|---|---|
| 1 month | $54.99 | $49.99 |
| 3 months | $164.97 | $49.99 |
| 6 months | $329.94 | $49.99 |
The Core Difference
LockedIn AI has built a capable desktop product with stealth claims. But “we are invisible” stated in marketing is a different thing from SetWindowDisplayAffinity(hwnd, WDA_EXCLUDEFROMCAPTURE) called at the Windows compositor level.
The fact that LockedIn AI advises users to rename their app executable in their own support documentation is a signal: the stealth is not architecture-deep. It relies on the process name not being recognized. That's a different class of solution than a named Win32 API that instructs the OS to exclude the window from all capture output regardless of what any observer looks for.
faFAANG's stealth doesn't require renaming anything. It doesn't require any user configuration. It is applied at window creation and maintained by the Windows Display Infrastructure. That is the difference between a verifiable guarantee and a workaround.
Stealth that names a Win32 API is a guarantee. Stealth that recommends renaming your exe is a workaround. faFAANG builds on the guarantee.