FriedrichAI: Offline AI

Norton users: Please make sure Norton is fully up to date before installing or running FriedrichAI. Older virus definitions or program versions may incorrectly flag parts of the app or block features such as local audio and microphone access.

AMD Support — Beta

AMD GPU support is available through a separate Radeon-compatible build and is currently in beta. Compatibility and media-generation performance may vary by GPU and driver. NVIDIA remains the most extensively tested GPU path.

Demo Available

A free demo is available for FriedrichAI.

The demo is designed to help you confirm that FriedrichAI launches and runs on your system before buying the full version.

It is a frozen compatibility snapshot, not a time-limited trial. It uses a conservative compatibility path, so performance may be slower than the full version on supported hardware.

If the demo does not launch or does not run well on your machine, please do not buy the full version yet. FriedrichAI runs locally, so hardware compatibility matters.

Run AI locally.

No required cloud. No subscription. No account.

Turn off Wi-Fi — FriedrichAI still works.

FriedrichAI is a private, self-contained AI workspace that runs directly on your PC.

Your chats, memory, saved conversations, local context, project data, and workspace information stay on your machine.

FriedrichAI is built for people who want AI running on their own hardware instead of someone else's servers.

Use it for writing, planning, problem-solving, brainstorming, project work, organization, research assistance, and everyday questions without depending on a cloud AI service for the core experience.

Most AI services rely on remote servers, accounts, recurring subscriptions, API billing, or usage limits.

FriedrichAI's core experience does not.

You own it. You run it. It answers to you.

Offline by default

FriedrichAI does not require an API key to run.

The main assistant, local memory, saved conversations, project tools, model management, and other core features are designed around local use.

Optional online and web-search features are available for users who choose to enable them. These are separate from the offline core experience and may require your own provider account, API key, or configuration.

If you only want a local offline assistant, you can ignore the online features completely.

Localized interface

FriedrichAI now includes interface localization for:

• English

• Ukrainian

• German

• French

• Spanish

• Brazilian Portuguese

• Korean

• Japanese

The interface language can be changed from Settings and is remembered between launches.

On first launch, FriedrichAI can automatically select a supported Windows language when available.

Built for local control

Because FriedrichAI runs on your own hardware, speed, model capability, and optional feature performance depend on your PC.

Local AI is different from cloud AI. A powerful system can run larger models and heavier features, while more modest hardware may need smaller models or CPU-friendly settings.

FriedrichAI supports multiple local execution paths depending on your hardware, including CPU, NVIDIA/CUDA, and Vulkan-supported configurations.

Start with the base assistant first. Confirm that local chat works before adding optional media packs, changing models, or adjusting advanced settings.

Local text-to-speech

FriedrichAI includes local text-to-speech support for spoken assistant responses.

Speech generation runs locally and does not require sending your conversation to a cloud speech service.

Text-to-speech can be disabled if you prefer a completely silent interface.

Optional media generation

FriedrichAI supports optional media features through additional downloads.

Depending on the installed pack and your hardware, these can include:

• Image generation

• Image variations

• Video generation

• Audio generation

Media generation is significantly more demanding than normal text chat and may require stronger hardware, additional storage, and large optional downloads.

NVIDIA currently has the strongest supported media-generation path.

AMD media-generation support is still under development and should not currently be treated as equivalent to the NVIDIA media stack.

This is separate from Vulkan/local chat support.

More than a chat window

FriedrichAI has grown into a broader local AI workspace.

Recent updates include:

• A cleaner desktop-style interface

• Saved local conversations

• Chat switching

• Saved-chat search

• Local project tools

• Project Board

• Project Chronicle

• Local memory

• Local model management

• Optional web search

• Local text-to-speech

• Interface localization

• Optional image, video, and audio tools

The goal is simple:

Chat, plan, save, organize, create, and keep working without handing your entire workflow to a cloud service.

Features

🔒 Local-first AI — the core assistant runs on your machine

📡 Offline by default — no cloud required for the main experience

⚡ No subscription required

👤 No account required

🔑 No API key required for the core assistant

🧠 Local memory — retain useful context on your own PC

💬 Saved chats — keep and reopen previous conversations

🔎 Saved-chat search — find older local conversations

🗂 Project Board — organize tasks into Todo, Doing, Blocked, and Done

📖 Project Chronicle — preserve useful local project history and context

🧩 Project tools — work through local projects and workflows with your assistant

🔁 Model Manager — scan and switch supported local GGUF models

🌍 Localized interface — English, Ukrainian, German, French, Spanish, Brazilian Portuguese, Korean, and Japanese

🔊 Local text-to-speech — hear assistant responses without using a cloud speech service

🖥 Runs directly on your own hardware

📦 Simple base setup — install, launch, and start chatting

🧰 Workspace tools for local files, notes, projects, and generated media

🎨 Optional media packs for image, video, and audio features

🌐 Optional web search for users who choose to enable it

What FriedrichAI can do

FriedrichAI can help you:

• Answer questions and think through problems

• Write, edit, brainstorm, and plan

• Work through ongoing personal or technical projects

• Organize tasks with the local Project Board

• Save and reopen separate conversations

• Search previous local chats

• Retain useful local context over time

• Work with local project files and notes

• Use supported local GGUF models

• Speak assistant responses locally

• Use optional web search when configured

• Generate images, video, and audio with supported optional media packs

• Continue using the core assistant without relying on a cloud AI service

Project Board

FriedrichAI includes a local Project Board for organizing work into:

Todo → Doing → Blocked → Done

It is a simple kanban-style system built directly into the app so you can break down ideas, track tasks, and keep project planning next to your AI assistant.

This is not a cloud project-management service.

Your board stays local to your machine.

Saved Chats

FriedrichAI supports persistent local chat sessions.

You can maintain separate conversations, reopen previous chats, switch between sessions, and search saved conversations from the interface.

This allows FriedrichAI to function more like a long-term local workspace instead of a disposable single chat window.

Local Memory

FriedrichAI can retain useful information locally across conversations.

Memory data remains on your machine and is designed to support longer-term work without requiring a remote account or cloud memory service.

You remain in control of the local environment where that information is stored.

Model Manager

The full version includes a local model manager for supported GGUF models.

Advanced users can place compatible models into the local model workspace, scan for them, and switch the active model.

The included default model works out of the box.

Model swapping is optional and intended for users who understand that different models can have very different RAM, VRAM, storage, and performance requirements.

Optional Web Search

FriedrichAI is offline by default.

Optional web-search tools are available for users who choose to enable them.

These features are separate from the core offline assistant and may require your own provider setup or API key.

The local assistant itself does not require web access.

Optional Media Packs

FriedrichAI supports additional media capabilities through optional downloadable packs.

These may include image, video, and audio generation tools.

Media generation can require significantly more storage and GPU resources than local text chat.

The optional media packages can also be large downloads, so they are not required for the base FriedrichAI experience.

Start with chat first. Add media only if you want it and your hardware can handle it.

Demo vs Full Version

The free demo is designed primarily to test baseline compatibility.

It is a frozen snapshot and may not contain the newest interface, localization, workspace features, model-management improvements, or optional systems available in the full version.

The full version is the actively developed FriedrichAI product and includes the main local assistant experience, local memory, saved chats, project tools, model management, ongoing updates, and access to supported optional feature packs.

The demo is not time-limited.

Important hardware note

FriedrichAI is local software.

That means your experience depends heavily on your machine.

A stronger PC can run larger models and more demanding optional features. A weaker system may need smaller models, CPU execution, or reduced expectations for generation speed.

Optional media generation has substantially higher hardware requirements than ordinary text chat.

If you are unsure whether FriedrichAI will run properly on your system, try the free demo before buying.

This isn't another AI service.

It is a local AI system you control.

No required cloud.

No required subscription.

No required account.

No required API key for the core local assistant.

Your AI. Your machine.



🔵 Portability

FriedrichAI is designed to remain self-contained and portable.

The application, models, chats, and local data can remain within FriedrichAI's own folder structure instead of depending on a remote account or cloud service.

This allows FriedrichAI to be moved, backed up, or run from an external SSD or USB drive on a compatible Windows PC.

Performance may be slower on slower external storage, but FriedrichAI and its local data remain under your control.

FriedrichAI uses Qwen models distributed under the Apache License 2.0.