01 / LOCAL
Work with your environment
Operate on Windows files, applications, local databases, PowerShell, and connected devices.
AGiENTEK Twin Agent
One AI system.
Two places to work.
A private-by-default Hermes deployment with one agent on your Windows desktop and one always-on in the cloud. They share memory through a protected channel—so your AI keeps moving without exposing your database or internal services to the public internet.
Persistent shared memoryLocal + cloud executionMobile access
Live system modelCoordinated
Local Agent
Your Windows workstationshared memory + context
Cloud Agent
Always-on VPS executionPrivate by default
Continuity without public exposure.
Keep sensitive, hands-on work close to your machine while the cloud agent handles uptime, messaging, scheduling, and background execution. Local ↔ VPS traffic stays behind a VPN or SSH tunnel.
Windows Local AgentDevice access, local UX, Windows actions, files, applications, and local workflows.
↔
VPS Cloud Agent24/7 execution, messaging, scheduling, and remote access behind authenticated ingress.
PostgreSQL + pgvector is the shared memory authority. Nginx is the only public application ingress.
Designed for real operations
Your agent does not stop when your laptop does.
Use the right execution environment for each job, without starting over or explaining the context twice.
02 / CLOUD
Keep work running
Run scheduled tasks, monitoring, web research, and automations from an always-on cloud agent.
03 / MEMORY
Compound intelligence
Shared PostgreSQL-backed memory gives both agents a common operational history and context.
04 / INGRESS
One controlled front door
Nginx is the only public application ingress. Databases, inference, proxies, and internal bridges stay private behind authenticated HTTPS.
05 / CONTROL
Own the infrastructure
Your deployment runs on infrastructure you control, with your provider keys, policies, and access boundaries.
06 / SUPPORT
Delivered, not DIY
AGiENTEK handles provisioning, configuration, security hardening, and end-to-end verification.
Deployment discipline
Secure. Stabilize. Verify.
The Twin Agent is built in the order that protects continuity: contain exposure, normalize services, back up shared memory, prove both-way recall, then optimize.
- Close direct public access to PostgreSQL, inference, proxies, and internal bridges.
- Keep Local ↔ VPS communication behind a private VPN or SSH tunnel.
- Back up PostgreSQL daily, store copies off-host, and rehearse restores.
- Verify Local → Cloud and Cloud → Local memory recall after restarts.
- Use FreeLLMAPI as a local-only routing proxy on each host.
- Use Ollama for embeddings unless generation has been separately validated.
Start small. Upgrade when ready.
Choose the level of reach you need.
Every tier is delivered by AGiENTEK. One-time setup gets you running; infrastructure, API usage, and optional ongoing support are recurring costs billed separately.
Solo Local
Private desktop AI
$97
one-time setup
Recurring: local operating costs and optional support
- Windows Hermes agent
- Local files, apps, and tools
- FreeLLMAPI configuration
- Handover documentation
Solo Cloud
Always-on cloud AI
$197
one-time setup
Recurring: VPS, API usage, and optional support
- 24/7 cloud Hermes agent
- Dashboard and gateway access
- FreeLLMAPI configuration
- Security and service setup
Twin Agent
One assistant, two homes
$497
one-time setup
Recurring: VPS, API usage, memory, and optional support
- Local + cloud Hermes agents
- Shared PostgreSQL memory
- Gateway and mobile access
- Monitoring and end-to-end verification
Ready when you are
Give your AI a second place to work.
Tell us what you want to automate and which systems your agent needs to reach. We will map the right Twin Agent deployment.
Book a Twin Agent briefing ↗Choose a 15-minute or 30-minute briefing time that works for you.