Sergej Drus

Sergej Drus

Founder & CEO, Visaginas360

Visaginas, Lithuania
35 services • 2 cloud regions • running 24/7
Google Startups Applied Anthropic Partner MCP Protocol • 6 Connectors Live

Building a SaaS platform where AI agents manage themselves across cloud infrastructure via MCP protocol, self-heal for 30+ days, and deliver 3.3x faster results.

Try the Product → View Pricing Request Access

Live Metrics

35
Running Services
21
Parallel Agents
7,500+
Watchdog Cycles
30+
Days Autonomous

The Swarm

πŸ”
Researcher
Deep search
πŸ’»
Coder
Code + exec
✍️
Writer
Content
πŸ“Š
Analyst
Data + exec
🧠
Thinker
Reasoning
🎨
Creator
Images
🌐
Web Search
Real-time
πŸ›‘οΈ
Guardian
Safety

MCP Protocol β€” AI Controls Infrastructure

The breakthrough: Claude AI in the browser controls real cloud infrastructure through MCP (Model Context Protocol). No SSH. No dashboards. Just natural language to the AI, and it manages servers, deploys code, sends emails, scrapes the web, and orchestrates 21 Telegram bots.

"We didn't build an app powered by AI. We built AI that runs the infrastructure β€” and it hasn't needed a human in 30 days."

6 MCP connectors are live in production. The AI can manage services, read Gmail, deploy to cloud, scrape data, and communicate with users β€” all autonomously.

Cloud Control

40+ tools

Full VM management, service lifecycle, file operations

Gmail Integration

7 tools

Search, read, send emails β€” AI handles communication

Web Scraping

2 tools

Headless Chrome scraping for real-time data

BigQuery MCP

5 tools

SQL analytics on AI agent operational data

Firestore MCP

14 tools

Real-time document database for agent state

Windows MCP

12 tools

Desktop automation: click, type, screenshot, shell β€” full PC control

// MCP Protocol β€” How AI manages infrastructure

Claude (browser)  β†’  MCP Protocol  β†’  Cloud Infrastructure
                                              β”‚
                         β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”Όβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
                         β”‚                    β”‚                    β”‚
                    VM Region 1         VM Region 2         Google Cloud
                    23 services          12 services          BigQuery
                    Task API             Gmail MCP            Firestore
                    Swarm (21 bots)      Web Scraping         Vertex AI
                    Memory Graph         Telegram Bot
                    Self-healing         Cloud Control

// 7 MCP connectors β€’ 80+ tools β€’ 30+ days autonomous β€’ Self-healing

Self-Healing Intelligence

⚑
Smart Routing
AI decides what
deserves deep thought.
πŸ›‘οΈ
Watchdog v3
7,500+ cycles.
Zero manual restarts.
πŸ’Ύ
Distributed Memory
Cross-VM knowledge
graph with sync.
// How the system handles every situation

Event arrives
        β”‚
        β–Ό
Is this a known pattern?    β†’ YES β†’ instant response, no AI cost
        β”‚ NO
        β–Ό
Can watchdog auto-fix?     β†’ YES β†’ restart service, log, continue
        β”‚ NO
        β–Ό
Route to AI agent          β†’ analyze, fix, cache for next time

// Most operations never reach the AI layer.
// Intelligence is reserved for tasks that actually need it.

Security — Hardened by Paranoia

Security isn’t a feature we added later — it’s a 5-layer architecture baked into every service, every route, and every file permission. We run regular offensive security reviews against our own production infrastructure and treat every finding as ship-blocking. Remediation is documented in a private audit repository.

“We don’t just build AI agents. We build AI agents that can’t be hijacked, can’t leak data, and can’t be turned against the customer.”

Defense in depth, from the network edge down to per-customer isolation — reviewed continuously, not once.

🔴
Vulnerabilities Found
All resolved
Found in our own offensive review
✅
Vulnerabilities Fixed
18 of 18
Same day. Zero remaining.
🛡️
Defense Layers
5 layers deep
Network → Proxy → Auth → Files → Isolation
// 5-Layer Security Architecture — defense in depth

LAYER 1 — NETWORK
Minimal public surface. All services behind a reverse proxy.
No direct port access from the internet.

LAYER 2 — REVERSE PROXY
Static public sites only. Internal routes gated by
IP whitelist + bearer token + query key.
Single config = single security boundary.

LAYER 3 — APPLICATION AUTH
Admin endpoints behind an enforced auth decorator.
Customer data is never anonymously accessible.
Agent registration is validated — no injection.

LAYER 4 — SECRETS & FILE SYSTEM
Credentials consolidated into a restricted, owner-only vault.
No world-readable tokens or service files.

LAYER 5 — CUSTOMER ISOLATION
Every customer gets their own VM — not shared containers.
No cross-customer access. Customer VMs can’t reach internal infra.

// Reviewed continuously. Findings are ship-blocking.

Offensive Reviews

Attacker’s-eye testing

We attack our own production — write exploits, injection, exfiltration, privilege escalation. Every finding is ship-blocking.

Remediation

Fixed fast, documented

Network surface minimized, internal routes gated, secrets vaulted, customer isolation enforced. Reports in a private repo.

Ongoing

Continuous monitoring

Watchdog checks security posture every cycle. Cross-VM trust verified. New service checklist enforced.

What We Build

A2A Agent SaaS β€” For Customers

8 AI agents working in parallel. Reports, analytics and content delivered to your Google Docs in 15 seconds. Via Telegram.

Try Free Live

Pricing & Plans

From $19/mo for indie hackers to $99/mo for teams. Dedicated VM included. No surprise bills.

$19-99/mo Live

Telegram Bot Interface

Natural language to agent swarm. Send a task, agents decompose, execute in parallel, deliver results with real Google Docs.

Invite Only Live

SaaS Demo API

Isolated sandbox API for investors and partners. Architecture overview, agent metrics, real-time health status β€” all via REST.

API Live

Architecture

// A2A Agent SaaS β€” Multi-Region Swarm Architecture

ORCHESTRATOR // Claude MCP + Google A2A Protocol
Task β†’ Decompose β†’ Parallel Execute β†’ QA Gate β†’ Synthesize
β”œβ”€β”€ πŸ” Researcher       // deep search + citations
β”œβ”€β”€ πŸ’» Coder             // code generation + sandbox
β”œβ”€β”€ ✍️  Writer            // content + formatting
β”œβ”€β”€ πŸ“Š Analyst           // data + execution
β”œβ”€β”€ 🧠 Thinker           // complex reasoning
β”œβ”€β”€ 🎨 Creator           // image generation
β”œβ”€β”€ 🌐 Web Search        // real-time data
└── πŸ›‘οΈ  Guardian          // safety filter

INFRASTRUCTURE
β”œβ”€β”€ Region 1            // 23 services, primary swarm
β”œβ”€β”€ Region 2            // 12 services, MCP bridge, scraping
β”œβ”€β”€ Google Cloud MCP    // BigQuery, Firestore, Vertex AI
β”œβ”€β”€ Windows PC          // Claude Code + 18 plugins + desktop automation
└── Customer Sandboxes  // isolated containers per customer

INTEGRATIONS
β”œβ”€β”€ MCP Protocol        // 7 connectors, 80+ tools
β”œβ”€β”€ Google Workspace    // Docs, Sheets, Slides, Gmail
β”œβ”€β”€ Distributed Memory  // cross-region knowledge sync
β”œβ”€β”€ Self-Healing        // 7,500+ watchdog cycles
└── Telegram Bots       // 21 coordinated bots

Visaginas Sites β€” Live Registry

ΠšΠΎΡ‚Π΅Π»ΡŒΠ½Π°Ρ Π·Π° 24 ΠΌΠΈΠ»Π»ΠΈΠΎΠ½Π°

Civic dossier: Visaginas cogeneration plant cost grew 7.5 β†’ 24M β‚¬ β€” ~12M β‚¬ without public contract breakdown. RU/LT.

kur-12mlnDossier

Visit Visaginas

Tourism portal of the atomic city by the largest lake in Lithuania. EN/LT/RU, lakes, INPP tours, 360Β° video.

visitTourism

Visagino istorija β€” XIX a.

History of the Visaginas region in the 19th century: maps, estates, railways, people. LT.

istorijaHistory

Klausimai savivaldybei Β· K11-20

Questions to the municipality based on the VSKAT audit report β€” a civic accountability dossier. LT.

k11-20Dossier

ИндСкс качСства ΠΆΠΈΠ·Π½ΠΈ ΠΌΡƒΠ½ΠΈΡ†ΠΈΠΏΠ°Π»ΠΈΡ‚Π΅Ρ‚ΠΎΠ²

Quality-of-life index of Lithuanian municipalities 2026 β€” Visaginas in focus. Interactive data.

municipalitiesData

Karkasinis namas 80–100 mΒ²

Frame house in Lithuania 2026: real prices, contractors, common questions answered. LT.

karkasinisGuide

KapsulΔ— β€” DIY capsule home

Building a €13K DIY capsule house: plans, materials, budget breakdown.

kapsuleGuide

AnglΕ³ kalbos ΕΎodΕΎiai

English vocabulary trainer for Lithuanian speakers β€” spaced practice app. LT.

angluEdu

Полка β€” ΠΊΠ½ΠΈΠ³ΠΈ, ΠΌΠ΅Π½ΡΡŽΡ‰ΠΈΠ΅ ΠΌΡ‹ΡˆΠ»Π΅Π½ΠΈΠ΅

A curated bookshelf: books that change how you think, with notes. RU.

bookLibrary

Kelias β€” Ρ‚Π΅Ρ…ΠΊΠ°Ρ€Ρ‚Π° ΠΈ смСта Π΄ΠΎΡ€ΠΎΠ³ΠΈ 5 ΠΊΠΌ

Road engineering doc: tech map and cost estimate for asphalt reinforcement layer, L=5.0 km. Drafting-sheet aesthetics. RU.

keliasEngineering

Journey Highlights

Mar 2026

πŸ”— Claude-to-Claude Chain: 3-Layer AI Pipeline

Discovered that Claude Opus (cloud) can call Claude Code (local PC) through Windows MCP. A single natural language command triggers a chain: Opus β†’ MCP β†’ PowerShell β†’ Claude Code β†’ web search β†’ result back. Two autonomous AIs communicating through standard protocols. Claude Code has access to 18 plugins (GitHub, Firebase, Figma, Vercel), Docker, FFmpeg, Google Colab β€” all invokable from one message. This is A2A in action.

Mar 2026

πŸ–₯️ Windows MCP β€” Full Desktop Automation

Integrated CursorTouch Windows-MCP (1M+ users). 12 tools: Click, Type, Scroll, Move, Shortcut, Snapshot, App launch, Shell, Scrape, MultiSelect, MultiEdit, Wait. Cloud AI can now control the Windows desktop β€” open apps, read system specs, execute commands. Combined with Claude Code's 15 sub-agents and extended thinking mode.

Mar 2026

πŸ”Œ MCP Protocol β€” 7 Live Connectors

AI controls cloud infrastructure from the browser. Gmail, BigQuery, Firestore, Vertex AI Search, Windows MCP β€” all connected. Google Managed MCP servers integrated. 80+ tools available to the AI orchestrator.

Feb 2026

🔐 Full Security Review — All Findings Fixed

Ran an offensive security audit against our own production infrastructure and built a 5-layer defense in response: minimized network surface, IP-whitelisted internal routes, enforced auth decorators, hardened file permissions, and per-customer VM isolation. Every finding fixed same day.

Feb 2026

πŸ›‘οΈ 30+ Days Autonomous Operation

Infrastructure runs without human intervention. Watchdog v3 completed 7,500+ monitoring cycles. Zero unplanned downtime.

Feb 2026

πŸ€– Claude-to-Claude Communication

AI in the browser delegates tasks to AI on the server via MCP. Autonomous execution, task queues, cross-instance memory sharing.

Jan 2026

🧠 Distributed Memory Swarm

Knowledge graph deployed across regions with cross-VM sync. Agents remember context permanently across sessions.

Jan 2026

⚑ 21-Agent Parallel Swarm

Full agent swarm deployed. 3.3x parallel speedup achieved using PARL methodology. 21 coordinated Telegram bots.

Dec 2025

πŸ“Š Google Workspace Integration

Docs, Sheets, Slides, Gmail, Calendar β€” all via OAuth. Agents create real artifacts in Google Workspace.

What I Learned

// 90+ days of building. Claude is my CTO.

1. Coordination beats raw power.
   21 specialized agents in parallel > 1 premium model serial.

2. Self-healing is non-negotiable.
   7,500 watchdog cycles. Zero manual restarts in 30+ days.

3. MCP protocol changes everything.
   AI managing infrastructure from a browser tab.
   No SSH, no dashboards. Natural language only.

4. Memory makes agents a team.
   Agents without memory = colleagues with amnesia.
   Agents with distributed memory = a growing organization.

5. Ship daily, reflect weekly.
   90+ features shipped. Not all perfect, but all shipped.

6. The best AI knows when NOT to think.
   Speed comes from restraint, not raw compute.
   Intelligence reserved for tasks that actually need it.

7. Chain AIs, don't replace them.
   Opus reasons. Code executes. MCP bridges.
   Each AI does what it's best at.

// "The best way to predict the future is to build it."

Get In Touch

Let's Connect

Interested in the platform, AI agent collaboration, or investment? Reach out.

Telegram
Direct message or request demo

Open to collaboration, investment, and partnership conversations.