The webAI Winter Release: Private AI for Everyone

The webAI Winter Release: Private AI for Everyone

At webAI, we believe AI should be owned by the user—scalable, powerful, and free from gatekeepers. Instead of relying on centralized, public cloud systems that demand you trade security for convenience, we’ve built a private, distributed platform that runs on hardware you already own. The result: full control of your data, flexible scaling, and the freedom to customize AI for your unique needs.

A Complete Private AI Stack

With this Winter Release, webAI is pleased to announce our end-to-end private AI ecosystem that streamlines everything from model training to global deployment. 

We’ve integrated our core technologies into a three-layer platform, each solving a different part of the AI puzzle but working together as one:

• Navigator: A local AI workspace to build, train, and deploy custom models.
• Infrastructure: A distributed network that scales AI worldwide, leveraging hardware you already own.
• Companion: A powerful assistant that lets users tap into a marketplace of private enterprise AI solutions—on any device.

This is the most complete private AI platform that’s ever been built. Below, we’ll dive into each layer and show why it’s time to take AI private.

Everything in the webAI Winter Release is available today, with a few exciting updates on the horizon. You can start using Navigator and Companion immediately to build, deploy, and access private AI solutions. Companion for iOS will roll out later in Q1, and the new Infrastructure capabilities including Team and Access Management will be fully available in February. 

Navigator: Your Local AI Workspace

From the beginning, we saw that relying on cloud servers to train and optimize AI models wasn’t just expensive—it also surrendered control of your data. Navigator was our answer: a dedicated “AI workbook” that resides on your hardware, giving you a self-contained environment to build, train, and customize models without ever sharing your data externally.

What’s New

In this Winter Release, Navigator has been completely redesigned for a smoother, more intuitive experience. We’ve combined data-building, model-training, and deployment tools into a single interface. You can fine-tune powerful open-source models for highly specific tasks, whether it’s drafting legal documents, analyzing financial spreadsheets, or assisting your software dev team.

Cluster management: We’ve introduced the ability to manage clusters directly within Navigator. Rather than manually juggling local devices and servers, you can now seamlessly add and orchestrate multiple compute nodes from a single interface.

Navigator automatically picks the fastest communication channel—Wi-Fi, Ethernet, or Thunderbolt—to keep your training jobs humming along.

webFrame: Another big leap is webFrame, our advanced backend for distributing large models across connected devices. webFrame applies dynamic optimizations so you can effectively run models like LLama, Qwen, Deepseek, Nemotron and more. In many cases, we see up to a 30% reduction in model size while retaining over 99% accuracy, making it feasible to run advanced AI on devices as modest as a MacBook Air.

This is powered by Entropy-Weighted Quantization (EWQ), a novel method that intelligently reduces model size by analyzing entropy distribution and selectively quantizing transformer blocks—maintaining high performance while significantly lowering memory requirements. We’ll share more about this breakthrough in an upcoming paper.

Scalable API Layer: We’ve pushed our API layer to support over 20,000+ concurrent requests per minute, meaning even a single laptop can serve multiple users with near real-time responses.

Organization Management: Easily share and manage access to devices and clusters within your team. Navigator lets you create organizations, assign members access to shared compute resources, and collaborate on model building and deployment—all from one interface. Simplify resource sharing while keeping everything private and under control.

Why It Matters

Navigator frees you from sending huge datasets and sensitive information to the cloud. Your data stays yours. You see exactly how your models are built and refined, and you have full control over who can use them once they’re trained. It’s AI creation on your own terms—scalable, optimized, and private.

Infrastructure: Deploy Globally, Operate Locally

It’s one thing to train a model on a single machine or small cluster, but what about organizations with multiple offices, data centers, or remote teams? Historically, scaling AI across locations has meant uploading massive models to a public cloud or purchasing specialized hardware that’s often underused.

This Winter Release introduces the webAI Infrastructure layer—a distributed private AI network that taps into hardware you already own. Think of it as a “private AI cloud” operating across your local devices and existing data centers, without any actual cloud dependency.

What’s New

• Compute Clusters: webAI Infrastructure unifies multiple nodes into a single pool of compute power. For example, a single rack with 13 Macs can provide hundreds of CPU and GPU cores—enough to run advanced AI families like Llama, Deepseek, Qwen, and more.
• Global Reach: Clusters in different locations can be linked to form a worldwide AI backbone, ensuring employees in various offices or remote sites all have low-latency access to the same private AI capabilities.
• Team and Access Management: Manage team access and permissions with Infrastructure’s lightweight tools. Administrators can control who can create and consume models and keep sensitive projects secure while enabling collaboration across distributed clusters and teams worldwide.
• Zero Cloud Dependency: Because the data and models are never offloaded to a third-party provider, security and compliance are significantly easier to maintain.

Why It Matters

You can now securely collaborate across teams and locations without compromising on privacy and control. Whether you’re a mid-sized startup or a multinational enterprise, by leveraging your internal devices—be they laptops, office clusters, or specialized on-prem servers—you can save on costs, avoid cloud lock-in, and keep data in-house.

Companion: A Secure AI Assistant

Once you’ve trained a model in Navigator and deployed it via Infrastructure, how do end-users actually access that AI day-to-day? Companion is our answer. Instead of requiring your employees (or customers) to log in to some external web service, Companion brings AI directly to their device—desktop, laptop, or even mobile phone.

What’s new

• On-Device: All processing happens locally or on your private network. No data leaves your control.
• Context-Aware: Companion integrates with your existing workflow—documents, spreadsheets, design tools, or sales software—so it can offer relevant, intelligent suggestions in real time.
• Custom Model Access: Users can tap into specialized models trained in Navigator with a single click or command. No more generic responses from a public LLM; it’s your AI, shaped by your knowledge base.

Practical Use Cases

Document Summaries: A 50-page contract doesn’t need to be a 2 hour reading project anymore. Drop it into Companion, and it churns out key clauses, renewal terms, and actionable insights in seconds.

‍

‍Sales Playbook: If your organization has a unique sales methodology, train a model in Navigator, deploy it to Infrastructure, and let sales reps ask real-time questions in Companion—“How do we position against company A, B, or C?”

Company Data Room: Searching through a massive folder? Ask Companion for “all contracts up for renewal in Q2” or “which design prototypes use Patent XYZ.” It processes the entire dataset locally, giving you quick answers without sending data externally.

Beyond Chat: Companion also helps you create new documents, compile email drafts, or integrate up-to-date info from the web privately—so you don’t waste time on manual research or the risk of public AI scraping your data.

Why It Matters

Companion is the user-facing layer that turns private AI from a theoretical capability into something genuinely helpful in day-to-day work. It breaks down barriers between AI models and the people who need them most, whether they’re sales reps, marketing teams, legal staff, or product designers.

One Unified Stack, Endless Possibilities

Navigator, Infrastructure, and Companion come together to form a private AI tech stack. Build specialized models, deploy them across a global network of devices, and deliver them to users seamlessly—all while keeping full control of your data and systems.

Ready to Get Started?

Explore the tools:

Navigator: Build and fine-tune models tailored to your needs.
Infrastructure: Deploy AI globally using your own hardware, with zero cloud reliance.
Companion: Empower employees with private, on-device AI that drives real productivity.

Start building now with Companion and Navigator
Download Navigator and Companion

Start planning company-wide deployments and special projects.

‍