How to Be More Ambitious With AI Models

Summary

Felix Rieseberg presents a practical framework for using AI models more ambitiously, based largely on his experience with Claude while stressing that the advice is not model-specific. He divides ambition into four dimensions: speed, altitude, depth, and range. Speed means completing familiar work faster; altitude means asking for the outcome beyond the immediate task, such as keeping an application crash-free or replacing a quarterly business review deck with a better decision-making process. Depth means pushing the model toward expert methods, raw data, demanding goals, adversarial checks, simulations, and independent validation rather than accepting polished but shallow output. Range means combining methods from multiple fields and assigning agents concrete research or production responsibilities instead of merely asking them to role-play specialists. The examples assume that models can access relevant data, install software, use computers, and operate through connectors or other tools. Rieseberg says higher-level workflows may reduce the need for humans to provide one task after another, but they require redesigned infrastructure and remain limited by what models can reliably do. He recommends asking what a leading professional would do, which raw materials and tools they would use, and what external evidence would establish correctness. The article connects this approach to self-efficacy and difficult goals, arguing that people must grant themselves permission to attempt work that recent model capabilities make newly plausible, while continuing to use outside validation and human judgment.