​

Abstract

This roadmap defines a four-phase plan for the AVM protocol, shifted to reflect current ground-level progress. Phase 1 covers foundational deliverables—MVP dashboard and beta SDK release. Subsequent phases focus on token launch, network decentralization, and scalable adoption, each mapped to the next quarterly milestones.

​

1. Introduction

AVM advances through a structured timeline to establish robust compute capabilities, expand functionality, decentralize network operations, and scale global adoption. The protocol’s MVP dashboard is imminent and the SDK is live for beta testers; the token launch is scheduled for Phase 2.

​

2. Development Phases

​

2.1 Phase 1: Foundation (Ongoing Q2 2025)

• Deploy AVM MVP with core execution capabilities and SDK available for beta testers.
• Launch the initial web dashboard for monitoring compute tasks (beta).
• Integrate with leading AI automation platforms via MCP (Model Context Protocol).
• Establish security, monitoring, and developer support infrastructure.

​

2.2 Phase 2: Expansion (Planned Q3 2025)

• Conduct $AVM token testnet and mainnet launch with initial distribution.
• Add multi-language support beyond Python (e.g., JavaScript, Rust).
• Release real-time analytics and reporting dashboards.
• Initiate ecosystem grants and strategic partnerships.

​

2.3 Phase 3: Decentralization (Planned Q4 2025)

• Transition to a permissionless, decentralized execution network.
• Enable tokenized incentives for compute providers under the MCP standard.
• Launch DAO governance modules for protocol parameter management.

​

2.4 Phase 4: Scaling (Planned Q1 2026)

• Execute full-scale marketing and adoption campaigns.
• Expand global node infrastructure for AI agent networks.
• Continuously optimize performance; introduce advanced features (GPU support, dynamic load balancing).

​

3. Strategic Development Plan

The initial AVM deployment (v0) is hosted on distributed cloud infrastructure, with an alpha dashboard for task monitoring. Future versions will incorporate a proprietary VM Solver, enabling peer-to-peer contributions of compute resources for LLM processing, enhancing network resilience and reducing centralization.

​

4. Developer Integration

Developers integrate AVM via:

1. High-Performance HTTP API — Direct LLM-to-AVM solver communication for efficient task execution.
2. Comprehensive SDKs — Available for TypeScript (@avm-ai/avm-vercel-ai) and MCP server (@avm-ai/avm-mcp), supporting Python code execution through the Vercel AI SDK and MCP-compatible clients.