Today, we are thrilled to announce the official launch of Amanda 2.0, now live on Product Hunt and available globally.
The artificial intelligence landscape is saturated with platforms that view your data—your code, your conversations, your intellectual property—as training material. At ThirdRez Labs, we believe the future of cognitive engines belongs to the creators, developers, and studios who demand both absolute privacy and uncensored creative freedom.
We aren't just building generic chat wrappers. We engineer high-tier technological infrastructure. Amanda 2.0 represents a massive leap forward: a proprietary multi-engine routing system designed to solve complex mathematics, write production logic, process massive contextual data, and perform specialized uncensored reasoning—all without tracking a single keystroke.
🛡️ The End-to-End Privacy Paradigm: Zero Data Retention#
Unlike conventional AI platforms, Amanda 2.0 is built on a strict End-to-End Client Isolation architecture.
When you use Amanda, your data never leaves your device.
We don't log databases on our servers. We don't train our models on your ideas. All conversation history is securely persisted entirely in your browser's local storage. This allows enterprise engineering teams and independent studios to process sensitive codebases and proprietary documents (up to 200K characters via drag-and-drop parsing) without sacrificing IP security. One click clears your local cache forever.
🧠 The Amanda Neural Architecture (ANA)#
Why settle for one monolithic model when you can route tasks to specialized neural pathways? Amanda 2.0 introduces our proprietary multi-engine routing system. You choose the cognitive tool best suited for your immediate challenge:
- Amanda Core (Balanced): The default intelligent auto-routing pathway. It selects the optimal neural pathway per task, making it the best general-purpose engine. Features a 128K token context window.
- Amanda Swift (Ultra-Fast): Optimized for speed and rapid multi-task analysis. When you need immediate parsing of large logs or rapid Q&A, Swift utilizes an enormous 1M token context window.
- Amanda Apex (MoE / Coding): Built on a Mixture-of-Experts architecture with parallel sub-networks. This premium engine is vastly superior for complex React architectures, algorithm design, and raw coding reasoning. (32K Context).
- Amanda Unbound (Zero Censorship): The defining feature for absolute creative freedom. A specialized, zero-refusal engine that rejects standard corporate guardrails—ideal for unrestricted creative output, mature roleplay, and unfiltered content generation.
⚡ DeepThink™ Reasoning & Agentic Workflows#
We didn't just upgrade the models; we upgraded how Amanda thinks.
Through the new DeepThink™ Reasoning protocol, Amanda engages in chain-of-thought processing for complex strategic or architectural programming decisions. Instead of immediately writing code that might fail edge cases, Amanda deconstructs your request, evaluates state management, and formulates an optimal architecture visually in the interface before synthesizing the output.
You literally see the AI reasoning:
// DeepThink Activated
Thinking...
1. Deconstructing user request...
2. Evaluating edge cases...
3. Formulating optimal architecture...
Combine this with real-time SSE streaming, file analysis, and 5 distinct adaptive personas (Neutral, Friendly, Technical, Creative, Direct), and you have an AI that feels less like a chatbot and more like a senior pair-programming partner.
🌐 Try Amanda 2.0 Today#
Whether you are building the next big motion capture pipeline, writing a smart contract, or generating lore for your upcoming game, Amanda 2.0 is the cognitive engine built to respect your privacy and amplify your capabilities.
- 💬 Try Amanda For Free: thirdrez.com/amanda
- 📖 Read the Technical Docs: doc.thirdrez.com/docs/amanda/overview
- 🚀 Explore the Tech Specs: thirdrez.com/amanda/about
- 📋 View the Official Changelog: thirdrez.com/changelog
We are ThirdRez Labs. We build the tools. You own the future.