SynthaMind

10-04-2025Go, TypeScript, ChromaDB, LLM, JSON, PostgreSQL

← Back to Projects

SynthaMind is an open-source research framework I've been developing to explore what's possible beyond dataset-bound AI — modular, recursive, and designed to grow.

Inspired by how real brains work — but designed from scratch for machines — SynthaMind introduces a system of autonomous cognitive agents that can reason, plan, recall, and adapt. Each agent is an isolated thinker, and together they form a synthetic mind capable of tackling novel tasks and evolving internal knowledge structures over time.


Why SynthaMind?

Most modern AI recycles answers. SynthaMind is about discovery.

  • A modular agent architecture where each part of the system thinks independently.
  • A cognitive loop that supports planning, memory, and reasoning.
  • A focus on recursive self-evolution, not static prediction.
  • Built for openness, transparency, and research collaboration.

How It Works

SynthaMind agents live in their own execution loops, receiving tasks and responding with insights. They communicate via structured JSON and are orchestrated by a central Task Manager. Memory is handled semantically via ChromaDB, enabling agents to recall past experiences.

The system is written in Go with optional TypeScript interfaces, and it’s built to scale from simple experiments to full autonomous cognition research.


My Role

I’ve been leading the design and development of SynthaMind from the ground up — including:

  • Agent lifecycle design (Init, Tick, Report)
  • Task manager orchestration
  • Modular memory interface with vector search
  • Multi-agent collaboration groundwork
  • Vision, roadmap, and research direction

SynthaMind is built as a foundation for a new approach to synthetic intelligence — one centered on structure, introspection, and emergent reasoning.


What’s Next

  1. Complete the multi-agent kernel
  2. Add agent monitoring and introspection tools
  3. Launch recursive task simulation experiments
  4. Expand to real-world data and symbolic abstraction

If you’re curious about AGI, AI autonomy, or just want to build something radically different — get in touch.

“The mind is not a monolith. It's a network of intentions, in constant negotiation.”

© 2025 Nenad Bursać. Licensed under AGPLv3.