Phase II marks FRAKTIΛ’s transition from early internal validation to public adoption and composability at scale. The platform opens to a wider audience of developers and system designers, introducing critical integrations, hosted agents and a flexible agent-as-a-service (AaaS) model. This phase focuses on expanding intelligence input, modular backend options and deeper collaboration between agents, paving the way for a decentralized developer ecosystem.
Phase II brings powerful improvements in coordination, verticalization and deployment, enabling teams to move from single-agent use cases to cooperative systems with shared context and dynamic workflows.
Key Capabilities Introduced
✦ Public Developer Access
The platform becomes available for open registration, allowing teams, DAOs and enterprises to build, test and deploy on FRAKTIΛ’s infrastructure.
✦ Model Layer Expansion
Support for external inference engines (e.g., GPT-4 via OpenAI, Claude via Anthropic or custom LLM endpoints). Agents can now select from multiple model backends.
✦ Agent-as-a-Service (AaaS)
Hosted agent instances that can be accessed via API endpoints or embedded as widgets into other applications. Includes uptime management, load balancing and real-time usage analytics.
✦ Add-On Marketplace Alpha
Launch of installable extensions:
➫ DeFi liquidity bots.
➫ CRM / email / scheduling automations.
➫ Voice UX modules.
➫ Analytics feeds. (on-chain + off-chain)
Add-ons are deployed as isolated microservices and managed via the FRAKTIΛ dashboard.
✦ Agent Teams v2
Support for shared memory, conditional triggers and multi-agent event flows. Enables parallel execution, branching logic and coordination across distributed environments.
✦ GPU Provider Integration
Connect agents to decentralized compute networks (e.g., Gensyn, io.net) for scalable inference workloads, reducing dependency on centralized APIs.
✦ Simulation Mode
Agents can now be tested in sandbox environments using Gazebo integration, ideal for pre-robotic or physical deployment testing.
Use Cases Unlocked
✦ Multi-agent workflows for DeFi arbitrage or portfolio automation.
✦ AI assistants embedded in Web2 interfaces, powered by hosted APIs.
✦ Swarm-based knowledge pipelines: Agent 1 pulls → Agent 2 filters → Agent 3 executes.
✦ Voice bots integrated in customer-facing apps with fallback to text-mode.
✦ On-chain agents that analyze wallet activity and trigger contract interactions.
Current Constraints
✦ Agent Teams still operate in off-chain coordination. (on-chain sync comes in Phase III)
✦ Marketplace is curated and limited to internal & whitelisted extensions.
✦ Compute layer abstraction is in early beta; provider variability may impact latency.
✦ Governance features (agent permissioning, DAO proposals) are not live yet.
✦ Hosted agents are not fully customizable via terminal yet. (read-only config initially)
Best Practices for Phase II Builders
✦ Choose model engines strategically
Fine-tune agent response time and cost using the most efficient API or local LLM integration per agent.
✦ Experiment with Teams
Create role-based micro-flows where each agent handles a specific stage: analysis, decision, action or logging.
✦ Monitor via AaaS dashboard
Hosted agents now offer observability: latency, uptime, voice activity or query volume.
✦ Test Add-Ons incrementally
Deploy 1–2 extensions at a time and validate runtime behavior using MCP diagnostics.
✦ Simulate before deployment
Use Gazebo for agents controlling robotics or external hardware logic to avoid real-world errors.
Strategic Role in Roadmap
Phase II scales the platform from tool to ecosystem, inviting external developers to build reusable modules, shared agents and plug-ins. It is the bridge between isolated deployments and composable, persistent infrastructure and sets the stage for full decentralization and DAO-based coordination in Phase III.