Two-Year AGI Development Plan

This roadmap outlines the specific technical and organizational goals for the next 24 months in the Hyperon and PRIMUS ecosystems.

Year 1: Foundation and Scalability

Q1-Q2: DAS Release & Stabilization

• Migration: Full transition of research projects from in-memory AtomSpace to the Distributed AtomSpace (DAS).
• Metric: Support for 10 billion Atoms with sub-second query latency across a 10-node cluster.

Q3-Q4: MeTTa Runtime Optimization

• JIT Compilation: Introduction of a Just-In-Time compiler for MeTTa rules to reach parity with native C++ execution speeds.
• Python-MeTTa 2.0: Deeper integration allowing automatic "Atomization" of Python objects and PyTorch models.

Year 2: Integration and Agency

Q1-Q2: Neuro-Symbolic "Grounding"

• Unified Percepts: Standardizing the PRIMUS PRUs to auto-generate MeTTa Atoms from real-time video feeds with 95% accuracy.
• Symbolic Control: First successful demonstration of a MeTTa reasoner controlling a complex robotic arm in an open-ended environment.

Q3-Q4: The First "Persistent Agent"

• Continual Learning: An agent that runs 24/7 in a simulation, building a memory of its environment and showing zero-shot adaptation to new objects.
• Metric: Scoring in the top 10% of the ARC (Abstraction and Reasoning Corpus) without task-specific fine-tuning.

Development Pipeline

gantt
    title AGI Engineering Roadmap
    dateFormat  YYYY-MM-DD
    section Scalability
    DAS Production Release    :2026-03-01, 180d
    100B Atom Support        :2026-09-01, 120d
    section Language
    MeTTa JIT Compiler       :2026-06-01, 240d
    section Integration
    PRIMUS Grounding Engine  :2027-01-01, 200d
    section Agency
    Persistent Simulation    :2027-06-01, 180d

Next: Challenges and Risks