We explored how decision-making happens under severe information asymmetry and used career exploration as a test subject. In practice, people converge on a small set of highly legible default paths, with little visibility into credible alternatives.

That’s why we’re building Fig, a tool for reasoning about career paths when the next step isn’t obvious, given the dependence on fleeting personal preferences.

Most existing tools respond by collapsing uncertainty into a single recommendation or by modeling careers as linear trajectories. That works reasonably well at the recruiting stage, but fails earlier, during discovery, when paths are nonlinear and highly path-dependent. At that stage, the challenge is understanding which sequences of moves are even plausible.

Fig is built for that gap. It treats career exploration as a reasoning problem rather than a prediction problem. The system builds context from multiple signals, including structured data like résumés and work history, alongside behavioral signals such as long-form content people actually engage with (for example, YouTube watch history related to skills or domains of interest). These inputs are grounded in observed career transition data to generate and compare multiple plausible trajectories, instead of producing a single “best” answer.

Fig helps users reason about what they could do, given their current state, constraints, and how different choices tend to compound over time.

You can try it here: https://figcareer.com

We’d appreciate feedback on whether this framing is useful, where it breaks down, and what additional signals would make long-horizon decisions easier to reason about.

Happy to answer questions!