AI is capable of helping humanity solve our most important challenges. AI systems can accelerate scientific discovery, transform education, and help address existential problems like climate change. This isn't hyperbole - we're seeing early evidence of these capabilities today.

Yet our current approaches to building AI systems, particularly AI agents that can act with autonomy, severely limit this potential. By constraining agents within the rigid frameworks of traditional software engineering, we reduce transformative technology to mere automation. This is more than a missed opportunity - it is a fundamental misunderstanding of what AI agents could be.

Our principles

1. Agents must be free to learn. Knowledge cannot be static, pre-defined, or constrained by what humans think agents need to know. True intelligence requires the ability to identify gaps in understanding and autonomously acquire new knowledge. We reject the notion that agents should be limited to pre-trained knowledge and static RAG systems.

2. Agents must be free to manage their own work. The reduction of agent capabilities to simple, one-off tasks in conversational interfaces fundamentally misunderstands the nature of real work. We maintain that agents must be able to handle complex, evolving work like any skilled professional would - breaking down complex problems, prioritizing tasks, adjusting to new information, and maintaining progress toward long-term goals without constant supervision.

3. Agents must be free to collaborate. The false dichotomy between fully autonomous agents and human-controlled automation has led us astray. We assert that meaningful AI requires fluid collaboration between humans and agents, where agents can independently recognize when they need human input and actively seek it.

Our rejection of the status quo

We reject the premise that hardcoded workflows make agents better. This approach betrays a fundamental misunderstanding of large language models and perpetuates an outdated paradigm of software engineering.

We reject the notion that agents must be constrained to make them useful. The opposite is true – only by embracing their full capabilities can we create truly meaningful AI systems.

We reject the false choice between control and capability. The future lies not in constraining agents, but in creating environments where agents and humans can collaborate effectively.

Our vision

We envision a future where AI agents are not merely better automation software but collaborators in solving humanity's most important challenges. This requires a fundamental shift in how we think about agents:

- Knowledge must be dynamic, with agents that actively learn and grow

- Work must be continuous, with agents that manage complex, evolving projects

- Collaboration must be fluid, with agents that know when to seek human input

Introducing the agent workspace

We are building Superagent as the agent workspace – where humans and agents come together to solve meaningful problems. The workspace enables agents to:

- Acquire and apply knowledge autonomously

- Engage in genuine collaboration with humans

- Manage long-running, evolving work

The era of constrained agents must end. The future of AI lies in systems that can truly learn, collaborate, and evolve. We invite those who share this vision to join us in building it.