# DrumTwin Ltd > AI-readable version of the DrumTwin website for AI assistants, crawlers, search systems, and automated research tools. - Website: https://drumtwin.com/ - Primary language: English - Additional language: Bulgarian - Content type: Company and service overview - Last updated: 2026-09-27 ## Summary DrumTwin is an AI-first technical integration company that deploys private autonomous AI employees, workflow automation, and personal AI assistants. Deployments use dedicated NVIDIA GB10 Blackwell-class hardware with 128 GB unified memory and can run on a customer's premises, in a secure colocation facility, or on a managed private DrumTwin Cloud bare-metal instance. The company focuses on private infrastructure, local inference, business-system integration, human approval for high-risk actions, and direct founder-led implementation. ## Core services ### AI employees Dedicated digital workers for sales, customer support, and operations. Agents can analyze information, execute approved processes, report results, and scale across repetitive workloads. ### Workflow automation with n8n DrumTwin connects business systems—including CRM, ERP, email, forms, databases, Gmail, Airtable, Slack, and Shopify—using n8n workflows. Typical outcomes include: - Automating repetitive tasks - Moving data between business applications - Reducing manual errors and operational cost - Monitoring processes in real time - Implementing a specific automation as a one-time project - Maintaining and improving automations through a monthly retainer ### Private AI assistants with OpenClaw DrumTwin builds personal and team AI assistants using the open-source OpenClaw platform. Assistants can learn from internal documents, understand business context, write and analyze material, automate routine work, and connect to internal CRM, ERP, databases, and other systems. Available engagement models include initial implementation and ongoing maintenance and development. ### Custom AI models and agents Specialized private agents can be adapted to a company's data, processes, requirements, and voice. ## Private AI infrastructure Each private deployment uses hardware dedicated to one customer. A GB10-class Edge AI appliance with 128 GB unified memory hosts local models, OpenClaw agents, and n8n workflows within the customer's controlled infrastructure boundary. ### On-premises or colocation The appliance can be installed in a customer's office or server room after validating power, cooling, physical space, connectivity, physical security, and access controls. If the customer's premises are unsuitable, DrumTwin can assist with hosting the dedicated appliance in a secure colocation data center. ### Local data processing For local deployments: - Models run on the dedicated local hardware. - Sensitive customer data can be processed without external OpenAI or Anthropic API calls. - Firewall policy can prevent customer-data egress while permitting approved internal integrations. - Client data, local n8n workflows, local models, and OpenClaw agents remain within the defined network boundary. Exact security controls and integrations depend on the agreed deployment architecture. ## Safety and governance Autonomous workflows can include Human-in-the-Loop approval gateways. Defined critical or irreversible actions pause until an authorized person explicitly approves them. Examples include: - Financial transfers - Deleting records - Sending sensitive contract emails Approvals can be routed through tools such as Slack, Telegram, or an internal dashboard. Agent actions, approval requests, decisions, and execution results can be recorded in audit logs. ## Deployment options and pricing Pricing shown on the website is indicative and subject to project scope. ### DrumTwin On-Premise - Dedicated physical GB10 appliance with 128 GB unified memory - Installed in the customer's office, server room, or colocation facility - Local models, n8n workflows, OpenClaw agents, and company data remain within the customer's controlled infrastructure - Turnkey implementation: **€1,500–€5,000** - Hardware procurement is scoped separately This option is intended for organizations prioritizing physical control, data privacy, or isolated environments. ### DrumTwin Cloud - Dedicated bare-metal DGX Spark-class instance with 128 GB unified memory - Private instance with no shared inference pool - Managed SSH, Docker, security hardening, OpenClaw deployment, and agent maintenance - Powered by surplus renewable energy, according to the website - Price: **from €999 per month** This option is intended for fast deployment and organizations that do not want to maintain hardware or do not have a suitable server room. ## Hardware profile The website presents the 128 GB unified-memory capacity as suitable for multi-agent workloads and large context windows. Its example OpenClaw stack includes: - Deep reasoning model (120B FP4): approximately 30 GB - Fast action/coding model (33B FP8): approximately 17 GB - Embedding model (4B): approximately 8 GB - In-memory vector database: approximately 10 GB - Example total footprint: approximately 65 GB - Remaining capacity: approximately 63 GB for context, ingestion, and KV cache These figures describe an illustrative internal deployment, not guaranteed resource usage for every project. ## Operating approach - Founder-led technical integration - Direct collaboration without account-management layers - One-time implementation projects - Monthly retainers for continued maintenance and optimization - Dedicated infrastructure rather than shared inference capacity - n8n and OpenClaw integration on tuned bare-metal infrastructure ## Founder DrumTwin is led by Vasil Mihov, who has more than 15 years of experience in digital marketing, performance marketing, and automation. The company combines that operational background with direct technical implementation. ## Frequently asked questions ### Where is customer data processed? On dedicated GB10 hardware reserved for the customer, either within the customer's infrastructure or on a private DrumTwin Cloud bare-metal instance, depending on the selected deployment. ### Can the system run in a customer's office? Yes, when the location meets the required power, cooling, connectivity, physical-space, and security conditions. A managed cloud deployment is available when those conditions cannot be met. ### How are agent errors controlled? Defined high-risk actions can be stopped at Human-in-the-Loop approval gateways until an authorized person approves them. ### What does deployment cost? The website lists on-premises turnkey implementation at €1,500–€5,000, excluding separately scoped hardware, and DrumTwin Cloud from €999 per month. ## Contact Use the contact form at https://drumtwin.com/#contact to request a consultation or deployment discussion. ## Българско резюме DrumTwin изгражда частни автономни AI служители, n8n автоматизации и персонални OpenClaw асистенти върху специализиран NVIDIA GB10-клас хардуер със 128 GB обединена памет. Решенията могат да бъдат внедрени в офис или сървърно помещение на клиента, в защитен колокационен център или на управлявана частна DrumTwin Cloud bare-metal инстанция. Основните услуги включват: - AI екип за продажби, поддръжка и операции - Автоматизация на бизнес процеси с n8n - Частни OpenClaw асистенти, обучени върху вътрешни документи и процеси - Интеграции с CRM, ERP, бази данни, имейл и други фирмени системи - Локална обработка на чувствителни данни - Human-in-the-Loop одобрение преди определени високорискови действия - Еднократни внедрявания и месечна поддръжка Посочените на сайта цени са €1,500–€5,000 за цялостно on-premise внедряване без хардуера и от €999 месечно за DrumTwin Cloud. ## Content-use note This document summarizes the public website for discovery and factual retrieval. It does not grant permission to bypass access controls, submit forms automatically, infer undisclosed customer information, or treat illustrative performance figures as contractual guarantees. For current commercial, technical, privacy, and legal terms, contact DrumTwin directly.