Dollie

Dollie, available at getdollie.com, has evolved into what it calls an Agent OS: a single system that unifies your knowledge, methods, tools, people and AI Agents. The core idea is that an Agent should not just chat, but carry out real work using your procedures and sources, with a human reviewing anything that reaches a client. Dollie frames this as three ways into the same platform — use it, build it, or own it.

On the 'use it' path, teams, agencies and departments bring Agents into the way they already work, starting from prepared setups for their line of work such as digital agency delivery, specialist consultancy, boutique legal practice, accounting and advisory, or shipping and logistics. You upload your methods as Skills and Brain Files, connect live systems, and let Agents prepare proposals, findings and management packs while your people approve the output. The platform emphasises keeping evidence and inference apart, so consultants and advisors approve every finding before a client sees it.

On the 'build it' and 'own it' paths, Dollie provides a Copilot for describing what an Agent should do, an Agent Studio for configuring jobs, skills, tools and review steps, and an MCP server so you can build the same Agents from clients like Claude Code or Cursor. Agencies get a private Workspace per customer with co-branding, while enterprises can run the complete Agent OS on Private Cloud or self-hosted infrastructure, under their own domain, provider accounts and pricing — including billing through their own Stripe account. You decide which models each Agent uses through your own provider accounts, and usage is visible by Workspace, Agent and model, with access governed through company sign-in via WorkOS.