We have seen artificial intelligence (AI) assistants that can generate content on your behalf, we have also seen autonomous AI agents that can think and act on your behalf, and we have seen vibe coding tools that can create entire apps and websites on your behalf. However, the most ambitious idea out of all is building agent-native companies.
Cofounder, the flagship product from the General Intelligence Company (GIC) of New York, is positioning itself as the operating system for a new class of business, one where a small founding team, or even a solo entrepreneur, can run every department through coordinated AI agents. GIC frames its mission as building the infrastructure for the one-person billion-dollar company, with agents handling engineering, sales, marketing, customer support, and ops while humans set direction.
- Cofounder was launched in September 2025 as a memory-driven automation system for individual founders handling email, calendar, document, and CRM workflows across connected tools.
- The company, founded by Andrew Pignanelli, announced an $8.7 million seed round in December 2025, led by Union Square Ventures with Acrew Capital participating.
With Cofounder 2, currently in research preview, the General Intelligence Company (GIC) is expanding that scope from assisting one operator to orchestrating an entire agent workforce organized on an internal org chart.
How Cofounder Actually Works
Cofounder is based on two main ideas: orchestration and memory. Instead of combining every interaction into a single chat window, the platform can organize specialized agents into an internal org chart.
- Each agent focuses on a specific area like engineering, sales, marketing, customer support, or operations, and works in its own dedicated workspace.
- A higher-level manager starts tasks, tracks progress, and coordinates work between departments, while the agents work simultaneously using tools like Linear, Slack, Notion, email, calendar, and CRMs.
What makes this system better than a glorified workflow builder is its memory architecture. Cofounder uses a layered design that combines working, core, and long-term memory, allowing agents to retain context over weeks and months rather than resetting session by session.
Key product capabilities the company highlights include:
- Department-level coverage: Cofounder 2 can orchestrate sales, marketing, customer support, and infrastructure agents. The product also includes workspaces for engineering and design work.
- Specialized workspaces per agent: Each agent can operate inside a function-specific environment rather than a single shared chat window.
- Plain-English automation: Users can describe what they want automated, and Cofounder will build the workflow and assign the right agents to run it.
- Always-on Knowledgebase: A persistent memory layer that pulls from email, calendar, CRM, and Notion to maintain the growing business context.
- Approval queues: Notifications surface any work that needs human sign-off before agents execute.
In Conclusion:
Pignanelli has been clear that GIC is betting on a structural shift in how companies get built. Whether or not Cofounder can genuinely allow a one-person billion-dollar company is an open empirical question because multi-agent reliability still falters on complex, judgment-heavy tasks across the industry. However, as artificial intelligence (AI) and agentic stack get better through 2026, Cofounder does stand out as one of the few products attempting to compress an entire company into orchestrated software, rather than ship another vertical agent. If you are planning to start a company, Cofounder could be the platform for you.
💡 For Partnership/Promotion on AI Tools Club, please check out our partnership page.
