Updates

We’re releasing the first version of OSI ( Orchestration System I ), the orchestration layer inside VANOS (Voice Agent Neural Orchestration System).

OSI gives you voice-first agents that can be trained on a single industry or company domain, then deployed for real support and communication use cases.

What you can do today with OSI:

• Role Based Agents: spin up an agent as a receptionist, scheduler, or support rep. Assign precise roles so agents operate with focus, increasing efficiency significantly.

• Custom Personalities: define tone, style, and behavior (professional, casual, empathetic, etc.).

• Identity & Voice Profiles: assign each agent distinct voices, names, and roles for clear customer interaction.

• Company Context: agents are tuned on your internal data/workflows so they respond with your company’s knowledge.

• Call Handling: fully functional inbound/outbound calls with conversation memory across a session.

Under the Hood:

• Single Agent Focus: each OSI agent is optimized for one company/industry domain at a time increasing efficiency significantly.

• Agent Kernel: modular layer that plugs in custom data, APIs, or CRMs to extend capability.

• Session Memory: retains context during calls (who the customer is, what was said earlier).

New in this Beta

• Deployment-ready voice agents for customer support and scheduling.

• Configurable personalities and voices out of the box.

• Customizable roles: receptionist, booking assistant, tech support, etc.

Next Miles

• Multi-agent orchestration (handoffs between roles: e.g., receptionist → sales → support).

• Cross-industry fine-tuning packs.

• Multi-lingual capabilities

• Real time analytics and supervisor dashboards.

We are building OSII, the second layer of VANOS (Voice Agent Neural Orchestration System). While OSI delivers focused, single-domain voice agents for support and calling, OSII expands into multi-agent orchestration with high concurrency.

What OSII enables
Multi-Agent Collaboration: Multiple voice agents can listen, converse, and coordinate in real time.
Parallel Conversations: Agents can run side by side, collecting and sharing information simultaneously.
Role Hand-offs: Customers can be seamlessly passed between specialized agents (e.g. receptionist → sales → support).
Shared Context: Information gathered by one agent is available to others in the same workflow.
High Concurrency: Thousands of agents can listen and talk at the same time, making it possible to handle operation-heavy communication loads at scale.

Important to note
OSII agents are orchestration-only, they do not directly execute external tasks. Their role is to converse, collect, and share information efficiently across large volumes of interactions.
The executional layer will be introduced in OSIII.

Under the Hood
Agent Graphs: Define workflows as networks of conversational roles.
Shared Memory Bus: Agents contribute to and access a common context layer.
Execution Scheduler: Coordinates thousands of concurrent conversational agents without bottlenecks.

New in this Alpha
Real time multi-agent voice orchestration.
Cross-role conversational memory.
High concurrency architecture for massive communication workloads.

Next Steps
OSIII: Executional agents capable of carrying out tasks.
Adaptive routing to automatically assign the right agent role.
Analytics for multi-agent communication flows.
Industry specific orchestration templates.

We’re very excited to share that work on OSIII has already begun. While OSI focused on single-domain conversational agents, and OSII enables multi-agent orchestration at high concurrency, OSIII will introduce the executional layer of VANOS.

OSIII is designed to move agents beyond conversation, enabling them to take real actions, not just listen and talk. This means:

• Agents that can both communicate and execute tasks in real time.

• Direct integrations with APIs, CRMs, and operational systems.

• Workflows where a voice agent not only gathers information but carries out instructions.

• A unified framework that combines orchestration (OSII) with action (OSIII).

Our executional strategy for OSIII has already been planned, and early prototypes are in motion. This is the step that transforms VANOS from a conversational orchestration layer into a true neural operating system for voice driven intelligence.