All your clients in one dashboard.
A Tier 2 AI engineer built right into the console.
No IDs, no passwords, no “please ask the client to start TeamViewer.” The PC is already connected — the technician clicks and gets to work.
CPU, RAM, disks, temperature, Windows events, RDP sessions — in one place, in real time, with no agent on the technician's laptop.
“Find out why 017-001 is running slow” — the AI connects, diagnoses and proposes a fix. The technician approves. Done.
Every action is audit-logged. Destructive commands are blocked by an AI reviewer (a senior model). Separate permissions for bulk operations and dangerous commands.
On top of the RDP session or in the client's system tray. Fully branded (your logo, your phone number). The client types — the technician sees it instantly in the console.
The same tool works on ten PCs or thousands. No need to switch stacks as your client base grows.
Recondeck is both a help desk admin console and a friendly tray icon on the client's PC. Each side gets exactly what it needs.
Ask the chat the way you'd ask a senior colleague. It connects to the right PC, reads the logs, finds the problem and proposes a fix. Runs on your own AI subscription.
# Pause Defender real-time protection for 30 minutes Set-MpPreference -DisableRealtimeMonitoring $true Start-Sleep -Seconds 1800 Set-MpPreference -DisableRealtimeMonitoring $false
Set-MpPreference -ScanAvgCPULoadFactor 25The AI picks the right model: a fast one for quick questions, a smart one for complex tasks, and a senior one for review.
Saves up to 80% of tokens on short requests.
Long conversation? The AI condenses older messages into a summary while keeping the key facts. The context never hits the limit.
Kicks in automatically when the context window is 70% full.
Before any dangerous command (wiping, deleting), a senior AI model must review it. If it's risky, the command is blocked until confirmed.
Protection against both AI mistakes and human error.
Every PC mentioned in an AI reply turns into an interactive chip:
🟢 017-001 CPU 12% RAM 4/8GBar / line / pie charts from any table. Mermaid for architecture diagrams. Process trees instead of raw JSON. Diffs for file edits.
The AI answers with visuals, not just text — like a real engineer at a whiteboard.
Every technician can tailor the look to their taste.
Real screenshots from the live console. Click any image to view it full size.
It's not just “have you tried restarting it?” The AI and the technician cover the full range — from everyday requests to serious engineering work.
The AI resolves most of these requests without escalation.
Half an hour of admin work → a couple of minutes in chat.
The AI reasons like a senior engineer — the technician signs off on the decisions.
One tool for your entire infrastructure — workstations, servers and cloud.
A full-featured remote administration console
WebRTC P2P, no relay. All-monitor overview, drag&drop file transfer.
Browse the agent's file system from the console. Upload, download and edit without RDP.
On top of the desktop or in the tray. Your own branding (logo, phone number).
CPU, RAM, disks, temperature, SMART, ping. History with weekly and monthly charts.
Disk cleanup, certificate renewal, printer setup — all with a single click.
Full disk, overheating, RDP intrusion — push notifications to the technician and to a Telegram chat.
Beta/prod channels, staged rollout, built-in kill switches. No manual reinstalls.
Tickets created automatically from alerts, with an ML priority classifier.
Agents keep a persistent connection to the server. The technician works in the browser through a local Bridge, which also calls the AI using your subscription. Remote desktop video travels directly, peer-to-peer, bypassing the server.
Minimal RAM (~30 MB), x86/x64, supports Windows 7 and later. No third-party dependencies.
Uses your existing AI service subscription — no extra per-request charges.
The video stream goes straight from the agent to the technician. The server only sets up the connection and never touches the traffic itself.
Essential when AI runs commands on someone else's PCs
format / del / Remove-Item -Recurse are blocked automatically on the server side. Running them requires explicit operator confirmation.
Before any dangerous command, a senior AI model assesses the risk and can block the command or suggest an alternative.
“Read-only,” “single agent,” “bulk operations,” “destructive” — separate permissions with an auto-expiry timer.
Every AI and operator action is stored in the database: who, what, when, on which agent, and with what result. Visible right in the console.
The server runs in your data center. Databases and backups stay with you. Full control over your infrastructure and your clients' data.
| Capability | Recondeck | LiteManager | Ammyy Admin | RustDesk |
|---|---|---|---|---|
| Built-in AI engineer | ✓ | ✗ | ✗ | ✗ |
| Bulk operations on 1,000+ PCs | ✓ | ✗ | ~ | ~ |
| Self-hosted (on-premises) | ✓ | ✗ | ~ (Enterprise) | ~ (Enterprise) |
| 24/7 metrics without an active session | ✓ | ✗ | ✗ | ~ |
| In-browser file manager | ✓ | ✗ | ✓ | ✓ |
| Unattended access (no client action needed) | ✓ | ✗ | ~ | ✓ |
Measured on similar tickets: “running slow,” “printer,” “disk full”
Leave your contact details — we'll get in touch and discuss exactly what you need.