4 minutes reading time
Introducing Freestate AI Companion—your always-on, always-free gateway to knowledge.
Freestate AI was created as a thoughtfully companion to support you through your personal and professional growth. Whether you’re a student, a creator, a curious mind, or a seasoned professional, this evolving ecosystem welcomes everyone.
What Makes Freestate AI Unique?
The mission: to add real value to the world by making knowledge more accessible, meaningful, and inspiring to all—no matter your age, background, or interests.
With Freestate AI Companion, you’re building a relationship with technology that personalizes itself to your learning style and goals, responds instantly to your questions and ideas, adapts quickly to how you engage, constantly evolving with you.
It’s not just smart—it’s supportive. It doesn’t just answer—it understands. And it’s completely free, because we believe real impact shouldn’t come with a price tag.
freestate AI uses medium-term memory, it remembers when you start a new chat, but the memory is cleared as soon as the application is closed.
This product turns natural language into functionality, proving that AI-driven development is not just possible, but powerful. It’s software created by intent, not syntax. No code.
Was designed and deployed on a lean, scalable way to build software, using AI-techniques prompt engineering.
Powered by Google’s advanced API model integrations, this software skips traditional coding. It leverages AI to translate human language into working systems—cutting dev time and lowering the barrier to innovation.
freestate AI: Under the Hood
Platform used to build the application: aistudio.google.com (Premium account recommended for efficient workflow).
Initial build time estimation: 1200 hours. (Continuously updating).
- How it was built: Architecture Freestate AI is built on a Client-Side First architecture, prioritizing privacy, speed, and direct interaction. Unlike traditional apps that route data through opaque intermediate servers, this application lives entirely in your browser.
- The Core: The application is written in TypeScript, ensuring strict code safety and reliability.
- The Brain: It connects directly to the Google GenAI SDK API, specifically utilizing the gemini-2.5-flash model for high-speed reasoning and low latency.
- The Memory: I utilized IndexedDB, a powerful browser-based database system. This means conversation history, uploaded files, and settings are stored locally on your device. No one ever sees your data; it remains yours.
- The Interface: The user interface is a reactive environment that updates at 60 frames per second using requestAnimationFrame, ensuring that text streaming, thought processes, and visual updates feel instant and fluid.
- How it works: Workflow when you send a message
- Context Assembly: Collects your current prompt, relevant history, and any active file attachments.
- System Prompt Injection: Injects a “System Prompt” – a strict set of internal behavioral rules written to ensure the system acts as a proactive and structured assistant.
- Stream Processing: Generates the response in real-time. The app uses a “Smart Diff” engine to parse the output into two streams: the Internal Thought Process (visible to you) and the Final Response.
- Action Execution: If the task requires it, it automatically triggers tools like Google Search for real-time data or the Deep Research agent for complex planning.
- Capabilities:
- Thought Process: Unlike standard chatbots, freestate AI displays its step-by-step reasoning logic before answering, ensuring transparency in how final answers are reached.
- Deep Research Mode: For complex queries, it generates a multi-step research plan, executing web searches for each step and synthesizing a complete report.
- Multimodal Analysis:
- Files: Can read and analyze PDF, DOCX, Excel, and code files.
- Vision: Can see and interpret images.
- Visual Generation: Can write code to render flowcharts, mind maps, and Gantt charts (via Mermaid.js) directly in the chat. These can be downloaded afterward.
- Voice Interaction: The app supports Push-to-Talk input and can read responses aloud using synthesized speech.
- Coding Assistant: Offers syntax-highlighted code solutions with debugging and refactoring capabilities. Uses the Google gemini-2.5-flash model.
- YouTube: Can extract and analyze video transcripts to summarize content or answer specific questions about a video.
- Limitations:
- Context Window: Although large, the “memory” regarding the current conversation is finite. Extremely long discussions may eventually lose the oldest details, so it is recommended to reintroduce context into the discussion.
- Browser Dependency: Since data is stored locally, clearing your browser cache/data will wipe the conversation history (unless you export it).
- Real-time Audio: While it can process voice input and output (transcribed to text), it does not support native, real-time voice conversations (like a phone call); it relies on ‘free’ text-to-speech engines.
Analogy:
The Lab (The App): This is the workspace. It has infinite whiteboards (Chat), filing cabinets (Storage), and specialized equipment (Deep Research Mode). It is secure; only you have the key.
The Assistant (freestate AI): This is your tireless lab partner. If you give it a stack of papers (PDFs, Excel sheets), it reads them instantly and provides summaries or other requested response formats.
– If you ask a complex question about the universe, it doesn’t just guess; it goes to the library (Google Online Search), verifies the information, and returns with a report citing the sources where it obtained the information.
– If you need to visualize a concept, it can immediately draw a plan in Diagrams, which can be downloaded later.
– The “Smart Lab” Think of freestate AI as a private, high-tech laboratory, where you are the scientist behind it all.
You can also find freestate.ro and freestate AI on: ‘YouTube‘, ‘Facebook‘, and ‘Instagram‘.