Videoglaz · Russia
Videoglaz — voice AI agent for business and video surveillance
A voice assistant on the company's own servers: data, stock, cameras, reports and tasks in one phrase.
Our platform · Russia
All casesNext Robot Local AI — on-premise AI server for the office
A compact AI server with 50+ agents and a gateway to 100+ neural networks inside the company network.
// Results
- digital agents in the pool
- 50+
- neural networks through one gateway
- 100+
- employees working at once
- 25
// How it works
Hardware for local inference: RTX 5090, Core Ultra 9, ≈3 L case 
One gateway to 100+ local and cloud neural networks 
50+ AI agents for legal, finance, engineering, HR and sales
What we built
We designed an all-inclusive platform: a ≈3 L server with an RTX 5090 (24 GB GDDR7) and Intel Core Ultra 9, local Qwen, Llama, Gemma and Whisper models, and a request balancer with queues, quotas and roles. External models connect by policy through a secure gateway that masks data. We deploy the knowledge base with RAG and OCR, set up agent templates for office processes and train the team.
- 50+ agents: legal, finance, engineering, HR, sales
- Gateway to 100+ local and cloud neural networks
- Balancer with queues, priorities, quotas and roles
- RAG on Qdrant with a reranker and source links
// Stack
- RTX 5090
- Local LLM
- RAG
- GLM-OCR
- Whisper
// Next case
// Your project
Let's build this for you
Describe the task. An engineer replies within 24 hours with a plan and an estimate.
- Free AI audit
- Fixed-price pilot
- NDA on request