Local-First Architecture
Oakden is built on a local-first architecture. This means that whenever possible, AI models run on your hardware, data stays on your machine, and cloud services are used only as an optional fallback. This is a fundamental design decision that affects everything we build.
What "Local-First" Means
In a traditional cloud-based AI product, every request you make travels over the internet to a remote server, gets processed there, and the result is sent back. Your data lives on someone else's computer. You pay for every query. If the internet goes down, the product stops working.
In Oakden's local-first model:
- AI inference runs on your hardware — Models are downloaded once and run locally using your CPU, GPU, or NPU (Neural Processing Unit)
- Data stays on your machine — Files, project data, and conversation history never leave your network unless you choose to use cloud features
- Cloud is a fallback — For tasks that exceed local hardware capabilities, Oakden can route to cloud providers, but this is opt-in and transparent
Why Local-First Matters
Privacy Your data never leaves your control. There is no third-party server storing your project files, home diagnostics, or business documents. This is especially important for construction companies handling contracts, financial data, and client information.
Speed Local inference eliminates network round-trips. Responses come back in milliseconds rather than seconds. For features like voice capture, real-time chat, and document scanning, this difference is significant.
Cost Cloud AI charges per token, per query, or per minute. Over time, this adds up to thousands of dollars. Local inference has a one-time hardware cost and then runs at the cost of electricity. For heavy AI users, the savings are substantial.
Reliability Local-first means your tools work even when the internet does not. Job sites often have poor connectivity. Rural properties may have intermittent service. With Oakden, your AI tools keep working regardless.
How the Oakden Router Works
The Oakden Router is the intelligence layer that decides where each AI request should go. It evaluates the complexity of the request, your hardware capabilities, and your routing preferences, then sends the request to the optimal provider:
- Local models (via Ollama) for most day-to-day tasks
- Cloud models (via OpenAI or other providers) for complex reasoning or vision tasks that exceed local capabilities
You control the routing through Eco Mode settings. See AI Routing & Eco Mode for details.
For End Users
If you are using Oakden products through their web interfaces (oakdenproperty.com, proaibuilder.com), local-first architecture is invisible to you. The products work like any other web application. The local-first infrastructure runs behind the scenes, and you benefit from the privacy, speed, and cost advantages without needing to configure anything.
For Organizations
Organizations that want maximum control can deploy Oakden products on their own hardware. This means all AI processing, data storage, and network traffic stays within your organization's infrastructure. Contact us for deployment options.