Comprehensive operations hub built on Next.js 16 and InstantDB, featuring live project dashboards, 7 Claude-driven agents, task management with server actions, and settings center for API keys and integrations.
Complete redesign with glassmorphism, improved information architecture, and comprehensive AI agent catalog. Enhanced UX with real-time collaboration and streaming agent results.
AgencyOS is an operational framework designed for running modern AI-powered agencies and consultancies. It provides structured processes, automated workflows, and methodologies for client work, project management, and service delivery. Built from real-world agency experience, AgencyOS helps organize operations into systematic, scalable processes.
Running an agency involves juggling multiple clients, projects, and deliverables while maintaining quality and profitability. Traditional project management tools are either too complex for small teams or too simplistic for sophisticated service delivery. Agencies need a middle ground: structured enough to ensure consistency, flexible enough to adapt to different client needs, and automated enough to scale without proportional overhead.
AgencyOS provides a complete operational infrastructure that codifies best practices into reusable systems. Built as a Next.js 16 console with InstantDB backend, it combines task/project tracking, AI agent catalog, and integration management into one unified interface. The platform leverages Claude Agent SDK for 7 prebuilt roles while providing typed server actions and real-time data subscriptions for seamless task management.
Built on Next.js 16 with App Router and Turbopack, AgencyOS uses InstantDB for real-time data synchronization with in-memory fallback for development. The dashboard provides live metrics via server components and InstantDB subscriptions. Task management uses typed server actions for create/update/bulk operations. The agents hub integrates Claude Agent SDK with internal tool registry for extensible automation. Health endpoints report database, SDK, and monitoring readiness for operational confidence.
The platform features a catalog of 7 prebuilt Claude-driven agents accessible through a unified hub: Research (market analysis), SWOT (strategic planning), Content (copywriting), CRM (client management), Planning (project roadmaps), Resources (asset discovery), and Outreach (communications). Each agent streams results in real-time via the Claude Agent SDK with built-in tool registry for extensibility. Outputs can be saved, edited, and integrated directly into client deliverables.
The dashboard surfaces live project/task metrics with quick actions and driver status from InstantDB subscriptions. Workspace and Tasks views provide full CRUD operations via typed server actions, supporting bulk updates and advanced filtering. Tasks link to projects for context retention while agents handle research, outreach, and content—reducing context switching for agency teams working across multiple clients simultaneously.
AgencyOS has been refined through real client implementations. The TileMart project serves as a prime example of AgencyOS in action: structured onboarding processes ensured clear requirements gathering, project templates provided consistent development workflows, and automated status reporting kept the client informed without manual overhead. The framework's flexibility allowed customization for TileMart's specific needs while maintaining operational consistency.
AgencyOS leverages Claude Code Resources to automate critical workflows. The /client-new command generates complete client onboarding structures, /project-status produces comprehensive status reports, and /email-draft creates client communications matching the agency's voice. This automation significantly reduces administrative overhead, allowing focus on high-value client work rather than operational tasks.
AgencyOS follows three core principles: Structure Without Rigidity (frameworks guide but don't constrain), Automation With Oversight (AI handles repetition, humans ensure quality), and Documentation As Code (processes are versioned, testable, and improvable). The system is designed to grow with the agency, starting simple for solo operations and scaling to support multiple team members and complex client portfolios.
Health endpoints expose database connectivity, Agent SDK readiness, and monitoring status for ops/CI validation. The platform includes demo mode with mock data when InstantDB isn't configured—enabling product walkthroughs and new developer onboarding without infrastructure blockers. Retry policies and driver fallbacks ensure the console remains usable even when external services degrade. This operational design makes AgencyOS production-ready while maintaining developer productivity.
AgencyOS provides a single pane for running client work: tasks stay linked to projects while AI agents handle research, outreach, and content generation. Teams eliminate context switching by managing all client operations in one console. The agent catalog and integration configs are declarative—adding new roles or connectors requires no UI rewiring. Real-time subscriptions keep team members synchronized, while typed server actions ensure data integrity across bulk operations.
AgencyOS continues to evolve based on real-world usage. Planned enhancements include enhanced financial tracking and profitability analysis, expanded integration with Claude Code for more automation opportunities, client portal functionality for self-service status updates, and a comprehensive service delivery playbook covering more specialized client scenarios. The framework is designed to be open-sourced, allowing other agencies to benefit from and contribute to its development.