↑

// SERVICE SPECIFICATION 04

Agentic AI Assistants - Hermes Agent

Deploying Hermes Agent, the open-source agent from Nous Research, as a self-hosted assistant that remembers across sessions, builds its own skills and runs scheduled tasks on your own infrastructure.

Deploy Hermes Agent Systems →

1. Service Overview & Scope

  • Self-Hosted Hermes Agent: Installing and configuring Hermes Agent (MIT-licensed, released by Nous Research in February 2026) as an always-on service on your Linux server or workstation.
  • Persistent Memory: Setting up session history and memory so the agent keeps project context, preferences and past decisions across conversations.
  • Skills From Experience: Configuring how the agent saves solved procedures as reusable skills, and reviewing those skills before your team relies on them.
  • Model Choice: Connecting Hermes Agent to local models through Ollama, llama.cpp, vLLM or LM Studio, to Claude, or to OpenRouter, depending on your cost and privacy needs.
  • Tools, Schedules and Access: Connecting MCP servers and internal tools, setting up scheduled tasks, and reaching the agent from the terminal or your team chat, all inside Docker with limited permissions.

2. Standard Architecture & Tools

  • Agent Framework: Hermes Agent with its memory store, skills library and task scheduler.
  • Model Backends: Local open-weight models (Qwen, Gemma, Llama, Mistral) via Ollama, llama.cpp, vLLM or LM Studio, plus Claude or OpenRouter through an OpenAI-compatible endpoint.
  • Tools and Integrations: MCP servers in Python or Rust, Python tool scripts, and messaging channels such as Slack, Discord or Telegram.
  • Isolation: Docker containers running as non-root, with limited network access and file permissions.

3. Deliverable Examples & 4-Week Schedule

  • Containerized Hermes Agent Deployment: Docker Compose setup with configuration, secrets handling, persistent storage for memory and skills, and backups.
  • Custom Tool & MCP Integrations: Python or Rust tools that let the agent query your databases, read internal documents and call internal APIs.
  • Starter Skills Library: A reviewed set of skills for your team's recurring tasks, written and tested before handover.
  • Evaluation Suite: pytest checks covering tool selection, argument correctness, memory recall and recovery from failed steps.
Timeline Focus & Architecture Core Deliverables & Artifacts Verification & Sign-off Criteria
Week 1 Hermes Agent Installation & Model Backend Configuration Docker Compose deployment of Hermes Agent on Linux server or workstation, persistent storage volumes, and model backend connection (Ollama, llama.cpp, vLLM, LM Studio, Claude, or OpenRouter) Always-on agent service responding reliably with chosen local or API model backend and baseline latency verified
Week 2 Persistent Memory & Task Scheduler Setup Session history configuration, cross-conversation memory store, scheduled task routines, and terminal/team chat messaging integration (Slack, Discord, or Telegram) Agent retains project context and preferences across distinct sessions and executes scheduled background tasks on time
Week 3 Custom Tools, MCP Servers & Skills Library Python and Rust MCP servers, internal database and document tools, schema-checked tool validation, and starter library of reviewed, tested skills for recurring procedures Agent executes multi-step tool calls with schema validation and successfully saves and reuses reviewed procedure skills
Week 4 Evaluation Suite, Docker Isolation & Handover pytest test suite covering tool selection, argument correctness, and memory recall, non-root Docker security lockdown, complete configuration transfer, and 30-day support onboarding All pytest evaluation checks pass, backup routines verified, complete configuration and IP handed over with client sign-off

4. Guarantees & Commitments

  • Private Operation Option: When paired with local models, the agent runs without calls to commercial LLM APIs.
  • Checked Tool Calls: Tool inputs are validated against schemas, and failed calls are caught, retried or escalated.
  • 100% Client Code & Configuration Ownership: All configuration, tools, skills and documentation transfer to you.
  • 30-Day Engineering Support: Help with model swaps, runtime tuning and new tool integrations.

5. Investment & Pricing Quote

  • Modular Service Selection & Flexible Engagement: Clients can freely pick and choose their services, either individually or bundled under our comprehensive transformation programs:
    • Individual Service Pick (Personal / 1-on-1): $100 USD per service for 5 days of live training + 5 days of dedicated support (10 days total coverage per service). Strictly 1 person only.
    • Individual Service Pick (Business / Teams): $200 USD per service for 5 days of live training + 5 days of dedicated support (10 days total coverage per service). Covers up to 20 participants (additional seats at $100 USD per seat).
    • Comprehensive Business Training: $2,000 USD for 60 days (30 days live training + 30 days guidance and detailed guides and materials). Includes cohorts up to 20 participants. Extra seats at $100 USD per seat).
    • Comprehensive Personal Training: $1,000 USD for 30 days (15 days live training + 15 days guidance and detailed guides and materials). Strictly for 1 person only (private acceleration).
  • Global Purchasing Power Parity (PPP) Available for All Service Pricing:
    • Tier 1 (Standard / 100%): Individual $100 USD | Business $200 USD | Business 60d $2,000 USD | Personal 30d $1,000 USD (USA, UK, EU, Canada, Australia)
    • Tier 2 (Developing Nations / 25% Off): Individual $75 USD | Business $150 USD | Business 60d $1,500 USD | Personal 30d $750 USD (India, Brazil, Mexico, UAE, Poland)
    • Tier 3 & 4 (Lower-Middle/Low-Income & Startups / 50% Off): Individual $50 USD | Business $100 USD | Business 60d $1,000 USD | Personal 30d $500 USD
    • Tier 5 (Academia & Non-Profits / 75% Off): Individual $25 USD | Business $50 USD | Business 60d $500 USD | Personal 30d $250 USD
    • Tier 6 (Crisis-Affected & Under-Resourced Regions): Individual $10 USD | Business $20 USD | Custom Subsidized Relief / Micro-Rates
    • Calculate and lock in your exact tier rate live on the Pricing page.
  • Payment Schedule:
    • Zero upfront advance; billed in 5 equal milestone installments due only after verified instruction and runtime delivery.

6. Summary

  • Hermes Agent, Self-Hosted: Nous Research's open-source agent running on your infrastructure.
  • Strict 1-Month Delivery: 4-week program from installation to production tools and handover.
  • Memory and Skills: An agent that keeps context across sessions and reuses reviewed skills.
  • Flexible Model Backends: Local models for privacy and cost, Claude or OpenRouter for harder work.
  • 100% IP & Code Transfer: Full ownership, deployment files and 30-day support.
Commission Agentic Systems → Book a free intro call on LinkedIn ↗