Fully Local & Private

Intelligence That
Never Leaves
Your Device

SEEV is a local-first AI workspace that runs on your hardware with no cloud backend and no telemetry. Core chat and memory features stay private on-device, while web search and external providers remain optional.

Download for macOS
Installation Instructions: After downloading the DMG, before opening the app, you must open Terminal and run sudo xattr -cr /Applications/SEEV.app. Enter your password when prompted to complete the authorization process and launch the application.
Explore Features
Zero Data sent to servers
100% Offline capable
Free No subscriptions
SEEV Workspace Interface
Core Capabilities

A Complete AI Workspace,
Built for Privacy

SEEV delivers a full suite of AI tools — code generation, document summarization, voice input, web research, and persistent memory — all running privately on your hardware.

Local AI Engine

Ships with a bundled local 1.2B model for first-launch use, plus optional 3B reasoning for tougher questions and attachments. Smart Hybrid routes between them locally.

Absolute Privacy

Your conversations never leave your device. No telemetry, no analytics, no data collection. SEEV has no backend servers to breach.

Offline Operation

Core chat, memory, and document workflows can run offline with the bundled local model. Connectivity is only needed for optional features such as web search or external providers.

Voice Input

Integrated WhisperKit speech recognition transcribes your voice locally. No audio is transmitted — voice data remains entirely on-device.

Code Generation

Generate production-quality code with syntax highlighting, language auto-detection, and one-click copy. Supports over 30 programming languages.

Workspace Anchoring

From chatting to building. Workspaces are the primary object. Drop PDFs and notes to tune the AI back to distinct domains instantly.

Editable Local Memory

SEEV keeps memory and uploaded document text local and user-editable, so saved context can be reviewed, updated, or cleared inside the app.

SEEV privacy architecture
Privacy Architecture

Privacy Is Not a Feature.
It Is the Architecture.

Unlike cloud-based AI assistants that process your data on remote servers, SEEV makes data leakage architecturally impossible. There are no servers to breach, no APIs to intercept, and no databases to compromise.

No Cloud Processing
Prompts and responses are generated entirely on your device. Nothing is transmitted externally.
No Account Required
No registration, no login, no email. Download and begin using immediately.
Zero Telemetry
No analytics, no tracking, no crash reporting to external services. Your usage is invisible.
Verifiable Architecture
Every component runs in your local environment. You can independently verify that no data leaves your machine.
Architecture

How SEEV Works

A zero-cloud architecture designed for performance and privacy. Every component runs locally on your hardware.

Data Flow — Complete On-Device Pipeline

Your Input
→
Local WebView
→
On-Device LFM
→
AI Response
→
Local Storage

Unlike cloud-based AI providers, SEEV keeps core AI flows on your device through a bundled app runtime and local storage. Optional web search and external-provider features use the network only when you choose them.

SEEV vs. Cloud-Based Alternatives

A direct comparison across key dimensions.

Dimension SEEV Cloud AI
Data residency On-device only Remote servers
Offline functionality Full offline Requires internet
Account requirement None Required
Subscription cost Free $10–50/month
Voice input (local) On-device WhisperKit Cloud-processed
Code generation Supported Supported
Data used for training Never Often by default
Persistent memory Local storage Cloud-synced
Workspace

Designed for Focused Work

A carefully crafted interface that balances power with simplicity. Projects, conversations, and quick actions — organized and accessible.

SEEV full workspace
Full workspace with sidebar, conversation view, and composer
Rich markdown response
Rich markdown rendering with headings, lists, and formatted text
Control Center settings
Control Center — model presets, persona, and inference settings
Structured table output
Structured data rendered as styled, readable tables
Specifications

Built on Modern Foundations

Engineered with cutting-edge technology to deliver fast, private, and capable AI on any modern Mac.

1.2B
Parameters in bundled model
WASM
Local app runtime
~1.2 GB
Local cache footprint
0 ms
Network latency

Ready to Own
Your Intelligence?

Download SEEV and experience AI that is truly yours. No accounts, no subscriptions, no compromises.

Download for macOS
Installation Instructions: After downloading the DMG, before opening the app, you must open Terminal and run sudo xattr -cr /Applications/SEEV.app. Enter your password when prompted to complete the authorization process and launch the application.
View All Features
Third-Party Notices

SEEV includes original SEEV code and interface elements together with third-party software, web libraries, model weights, and transcription assets. Third-party components remain owned by their respective copyright holders and are provided under their own licenses. Nothing in this notice transfers ownership of third-party software, models, trademarks, names, logos, or documentation to SEEV.

This notice applies to the SEEV website, bundled web application, app builds, and distributed packages that include the components listed here.

MIT BSD-3-Clause Apache-2.0 Liquid AI LFM Open License v1.0 Model Assets

General Notice

  • AI output limitations: SEEV may generate incorrect, incomplete, outdated, biased, unsafe, or misleading outputs. Users are responsible for reviewing outputs before relying on them.
  • No professional advice: SEEV does not replace a licensed professional, regulated advisor, emergency service, compliance officer, lawyer, doctor, accountant, financial advisor, or security reviewer.
  • Local storage: SEEV may store chats, projects, settings, memory entries, attachments, document retrieval snippets, model files, cached model data, voice transcription state, and related local state on the user's device.
  • Voice and transcription: If microphone or voice features are used, audio may be processed to generate text transcriptions. Transcriptions may be inaccurate and should be reviewed before use.
  • User responsibility: Users must not use SEEV, its outputs, or included models for illegal, harmful, infringing, deceptive, abusive, privacy-invasive, or restricted purposes.
  • No endorsement: Third-party licensors, model providers, open-source projects, package maintainers, and trademark owners do not sponsor, approve, certify, or endorse SEEV unless a written agreement says otherwise.

Runtime App Dependencies

  • WhisperKit from argmaxinc/WhisperKit, resolved version 0.17.0, is identified as MIT-licensed.
  • yyjson from ibireme/yyjson, resolved version 0.12.0, is identified as MIT-licensed.
  • swift-argument-parser 1.7.1, swift-asn1 1.6.0, swift-collections 1.4.1, swift-crypto 4.3.0, swift-jinja 2.3.5, and swift-transformers 1.1.9 are identified as Apache-2.0 components.
  • Apache-2.0 components require license text, copyright preservation, upstream NOTICE preservation where provided, and modification notices where applicable.

Bundled Web UI Libraries

  • marked is bundled as vendor/marked.min.js and is identified as MIT-licensed.
  • highlight.js is bundled as vendor/highlight.min.js, identified as version 11.9.0, and is BSD-3-Clause licensed.
  • highlight.js Atom One themes are bundled as atom-one-dark.min.css and atom-one-light.min.css; exact theme license should be verified before public release.
  • KaTeX is bundled as katex.min.js, katex.min.css, and KaTeX font files, identified as version 0.16.9 and MIT-licensed.
  • uuid is bundled as vendor/uuid.min.js and is identified as MIT-licensed; exact source/version should be verified before release.

Model Assets

  • Liquid AI LFM2.5 1.2B Instruct MLX 4-bit may be included under WebAssets/models/LFM2.5-1.2B-Instruct-MLX-4bit.
  • Liquid AI LFM2-VL 3B 4-bit may be included under WebAssets/models/LFM2-VL-3B-4bit.
  • OpenAI Whisper / Whisper large-v3 derived CoreML assets may be included under Sources/Chatbot/WhisperModels. Each converted or packaged model bundle should ship with exact provenance, model-card, and license metadata.

Liquid AI LFM Model Terms

Some SEEV features may use Liquid AI LFM model weights. Liquid AI LFM models are licensed separately by Liquid AI, not by SEEV. The LFM Open License v1.0 is based on Apache-2.0 but includes additional conditions for commercial use, attribution, termination, and trademarks.

  • Liquid AI retains ownership of the base LFM models. Derivative models remain subject to the applicable Liquid AI license.
  • Distribution of LFM models or derivative model files requires attribution, preservation of copyright/license notices, inclusion of the license, and marking modified files where applicable.
  • Current Liquid AI documentation describes free commercial use for organizations under $10 million USD in annual revenue; organizations above that threshold must obtain a commercial license from Liquid AI to continue commercial use.
  • The LFM Open License does not grant trademark rights except for reasonable use needed to describe model origin and reproduce required notices. Liquid AI and LFM names remain marks of their owner.

Commercial licensing contact identified by Liquid AI documentation: sales@liquid.ai. SEEV should include the full LFM Open License v1.0 text with any build that distributes LFM model weights.

Commercial Use, Distribution, and Release Requirements

  • MIT components generally permit commercial use, modification, distribution, sublicensing, and private use, provided copyright and permission notices are included.
  • BSD-style components generally permit commercial use, modification, distribution, and private use, provided copyright notices, license terms, and disclaimers are preserved and contributor names are not used for endorsement without permission.
  • Apache-2.0 components generally permit commercial use, modification, distribution, and private use, and include patent grants, but require license text, notice preservation, and modification notices where applicable.
  • Before public release, SEEV should include full license texts for MIT, BSD-3-Clause, Apache-2.0, and the applicable Liquid AI LFM Open License.
  • Before public release, SEEV should add exact source URL, version, license, model card, and provenance metadata for every bundled model file, and preserve upstream LICENSE and NOTICE files where provided.
  • SEEV should verify screenshots, background images, icons, design assets, generated images, and website assets before public distribution.
  • If distributed through the Mac App Store or another marketplace, privacy labels must match actual local storage, microphone, network, model, and telemetry behavior.

SEEV includes third-party software, model weights, web libraries, transcription assets, and related files that are licensed by their respective owners. These components are provided under separate open-source, source-available, or model licenses, including MIT, BSD-style, Apache-2.0, and the Liquid AI LFM Open License v1.0. SEEV does not claim ownership of third-party software or model weights except where expressly stated. Third-party licensors do not sponsor, approve, or endorse SEEV. Users and deployers are responsible for complying with all applicable third-party license terms, acceptable-use limits, attribution requirements, commercial-use thresholds, privacy obligations, and laws.