Pluely v1: A Stealth AI Overlay That Hides From Screen Shares, Now Closed Source

What Pluely v1 Actually Solves

Pluely v1 targets a narrow but real pain point: the moment in a live conversation where switching to a browser tab to ask an AI a question would cost you the room. Interviews, sales calls, standups, lectures, and live debugging all share this constraint. The project's answer is a single translucent overlay that sits on top of your desktop, not a separate window or tab. You can summon it with a global hotkey, ask a question about your screen, or have it listen to a conversation and suggest answers. The intended user is someone who needs AI assistance during synchronous interactions where visible tool-switching is socially or practically costly. The README is explicit about the use case: "the moments where switching to a browser tab would cost you the room." This is not a general-purpose assistant for writing code or drafting documents; it is a conversation-time aid with a specific stealth requirement baked into its design.

The Stealth Mechanism: How It Stays Invisible

The core technical claim is that Pluely is excluded from screen capture, absent from recordings and screenshots, and never steals focus from the active application. How this works on a technical level is not fully detailed in the README. It mentions an option to hide the icon from the Dock or taskbar, and it says the overlay is "invisible on screen shares." The mechanism likely relies on operating system APIs that allow a window to be marked as not capturable, such as macOS's screen capture exclusion or Windows' window enumeration filters. The README does not specify which APIs are used, and the source is closed, so independent verification is impossible from the repository alone. The claim that it "never steals focus" is plausible for an overlay that does not activate on its own, but the exact focus management is undocumented. What is clear is that the design prioritizes not appearing in any captured output, which is a different goal from simply being a small window. The README's phrase "no bot ever joins your calls" distinguishes it from meeting-bot solutions that join as participants.

Ask Mode and Listen Mode: The Two Workflows

Pluely v1 has two modes. Ask mode lets you type a question, dictate it with push-to-talk, capture a screenshot, drag-select a region, attach files, or enable an "Use image" option that sends a fresh screenshot with every message. Documents go through built-in OCR and remain in context for follow-up questions. Answers stream in as Markdown and are saved locally. Listen mode is more distinctive: you hit Start, and Pluely transcribes your microphone and system audio live, with speaker labels and language selection. It can fire automatic responses when someone asks a question, after a pause, or only when you tap a Suggest button. The README gives an example of a sales call where a suggested answer appears in real time. Both modes share a local SQLite database for chats, meetings, transcripts, and files. The distinction matters: Ask mode is a screen-aware Q&A tool, while Listen mode is a real-time conversation copilot. The latter is the feature that most directly competes with meeting-bot services, but it does so without joining the call.

Getting It Running: Downloads, Keys, and Providers

Pluely v1 is distributed as signed binaries for macOS (.dmg), Windows (.msi/.exe), and Linux (.deb/.rpm/.AppImage). There is no source build from the repository because the v1 source is closed. The README states that the free plan requires no account to start and updates ship automatically. For model access, you have two paths. The first is the hosted option on Pro plans: over 200 models including GPT, Gemini, Claude, and Llama, with no API keys to manage. The second is the free forever path: you connect your own LLM or speech-to-text provider through a curl template, or plug in AI CLIs you already have, such as Claude Code, Gemini CLI, Codex, Qwen Code, or Ollama. The README emphasizes that your own keys stay local and that conversations never train anything. The practical setup is therefore not a matter of compiling from source but of downloading a binary, choosing a provider path, and configuring keys. The documentation at docs.pluely.com includes a 5-minute quickstart, which is the recommended entry point.

The Closed Source Reversal: A Real Trade-off

The most significant fact about this project is that v1 is closed source, despite the repository being listed as GPL-3.0. The README explains: "Years of open Pluely code kept getting repackaged and sold as clones, and no license or complaint ever stopped it. So v1 ships as signed binaries while the product itself stays free to use at its core." This is a direct reversal from earlier versions that were open. The license on the repository is GPL-3.0, but that applies to whatever source is present, not to the distributed binaries. For an engineer evaluating this, the implication is that you cannot audit the stealth mechanisms, the local data handling, or the update process. The privacy claims rest on trust in the vendor. The README says data lives in a local SQLite database and provider keys stay local, but without source access, those claims cannot be independently verified. This is a genuine limitation for anyone who needs to confirm that the overlay is truly excluded from screen capture or that no telemetry is sent. The closed-source decision is understandable from a business perspective, but it changes the risk profile for adoption.

Licence and Maintenance Costs

The repository carries a GPL-3.0 license, but the v1 binaries are not distributed under that license in any meaningful way because the source is not available. The README does not specify a separate end-user license agreement for the binaries. For maintenance, the release history shows a gap: app-v0.1.9 was pushed on 2026-01-14, and app-v1.0.0 on 2026-07-14, roughly six months later. That suggests a substantial rebuild period. The README says updates ship automatically, which means users are dependent on the vendor's release cadence and continued willingness to maintain the product. There is no indication of a community contribution path since the source is closed. For an organization, the cost is not a license fee (the free plan exists) but the loss of control over updates and the inability to patch or fork the software. If the vendor stops maintaining it, there is no fallback. The GPL-3.0 tag on the repository is misleading to a casual observer, and that is worth noting in any evaluation.

Alternatives and What Pluely Is Not

The closest alternatives are meeting-bot assistants like Otter.ai or Fireflies.ai, which join calls as participants and provide transcripts and summaries. Pluely differs fundamentally: it does not join the call, so it leaves no trace in the participant list, but it also cannot capture audio from a call unless it has access to system audio. That is a key constraint. If you are on a video call where system audio is not accessible, or where the OS restricts audio capture, Listen mode may not work. Another alternative is a general-purpose AI desktop app like a browser-based ChatGPT or a local Ollama client, but those do not offer the same screen-capture exclusion or overlay behavior. Pluely's differentiation is the combination of overlay and stealth. The wrong tool case is when you need a collaborative transcript that other participants can see, or when you require full source transparency. For those needs, a meeting bot or a plain LLM client is more appropriate. The README does not mention any limitation regarding system audio access, but it is a reasonable inference that Listen mode depends on OS-level audio capture permissions.

Editorial conclusion

Adopt Pluely v1 if you need a fast, screen-share-invisible AI assistant for live conversations and are comfortable with a closed-source binary that stores data locally in SQLite. Do not adopt it if you require source access for auditing or customization, since v1 ships as signed binaries only. Before adopting, verify the stealth claims on your specific video conferencing tool and screen recording setup, and confirm that the free tier's own-keys approach covers the providers you use.

Official sources

1. iamsrikanthnani/pluely on GitHub
2. License: GPL-3.0
3. Project website
4. README
5. Releases