

Elephas
Privacy-first Mac AI writing assistant that redacts PII before any cloud call and indexes your files locally.
In short
Elephas is a native macOS app that protects client data by redacting 28 PII categories before cloud calls and indexing documents on-device. It is best for professionals under NDAs who need secure AI writing and research tools without uploading raw files.
Pick Elephas if you're a Mac-based consultant, lawyer, or analyst who wants Claude/GPT productivity without uploading raw client documents.
Skip it if you're on Windows or Linux, need a hosted team workspace, or want an open-source tool you can audit.
Elephas is a native macOS AI assistant pitched at professionals who can't pipe client data into a public chatbot. It indexes your documents on-device, redacts 28 categories of personally identifiable information before sending anything to a cloud model, and offers a fully offline mode for the most sensitive work. The headline features are 'Super Brain' (project-scoped knowledge workspaces built from PDFs, Word, Excel, Notion, Obsidian, Markdown, and Zoom transcripts) and 'Super Command' (a system-wide hotkey that drops AI writing and rewriting into any Mac app).
The target buyer is obvious: consultants, lawyers, accountants, financial advisors, and researchers who work under NDAs and want Claude or GPT without leaking names, SSNs, or account numbers. Elephas brings its own model bench (GPT-5, Claude, Gemini and 15+ providers via your own API key), so you can route confidential work to a self-hosted endpoint or a cheaper provider as needed. A 7-day Pro trial is available without a credit card; paid plans live on the pricing page.
Worth noting: this is a Mac-only desktop app, not a web product, and the Mac App Store build is deliberately crippled (no Super Command, no system-wide writing) because of Apple sandbox rules, so power users should install directly from the website.
Elephas is one of the few AI writing apps where the privacy story is more than a footer disclaimer: local indexing plus on-the-fly PII redaction is a credible answer for NDA-bound work. The Mac-only constraint and the gimped App Store build are real limits, but for the right professional it earns the subscription.
— The AI Tool Bible editorial team
Pros
- ✅ Automatic PII redaction across 28 categories before any cloud call
- ✅ Local indexing and a true offline mode for confidential work
- ✅ System-wide Super Command hotkey writes inside any Mac app
- ✅ Bring-your-own-key support for GPT-5, Claude, Gemini and 15+ providers
Cons
- ⚠️ macOS only; no Windows, Linux, or web client
- ⚠️ Mac App Store build omits Super Command and writing features
- ⚠️ No public API documented
- ⚠️ Closed source, so privacy claims rely on trust
Use cases
Frequently asked
- How does Elephas protect sensitive data?
- It redacts 28 categories of personally identifiable information before sending data to cloud models and indexes your documents locally on your device. It also offers a fully offline mode for the most sensitive work.
- Which AI models can I use with Elephas?
- Elephas supports GPT-5, Claude, Gemini, and over 15 other providers via your own API key. This allows you to route confidential work to self-hosted endpoints or different providers as needed.
- Is Elephas available on Windows or Linux?
- No, Elephas is a macOS-only desktop app. It is not available for Windows, Linux, or as a web client.
- What is the difference between the App Store and website versions?
- The Mac App Store build is limited by Apple sandbox rules and omits the Super Command hotkey and system-wide writing features. Power users should install directly from the website to access all features.
- Does Elephas offer a free trial?
- Yes, a 7-day Pro trial is available without requiring a credit card. Paid plans start at $19 for the Standard tier.
Explore related
Compare with similar tools
All in Writing →
Claude
FeaturedAnthropic's flagship assistant for long-form writing, analysis, and coding.

GPT-4o
FeaturedOpenAI's multimodal flagship behind ChatGPT.

Gemini Advanced
Google's flagship — strong at math, long context, and Workspace integration.

Llama 3
Meta's open-weights LLM family that put serious frontier-adjacent models in everyone's hands.

Qwen
Alibaba's open-weight foundation model family covering chat, vision, image generation, translation, and safety classification.

Jan
Open-source desktop ChatGPT alternative that runs local LLMs and routes to cloud providers from one app.