A solid electrolyte is not just a crystal lattice: it’s a bustling metropolis for ions. Each act of conduction resembles a trip through the city: there are wide highways with low energy barriers, gridlocks in bottlenecks, and tricky detours where an ion can get stuck. For decades, engineers, like clumsy city planners, randomly rearranged atomic ‘buildings,’ hoping to stumble upon the perfect transport network. Today, intuition is giving way to an autonomous planner — an alliance of machine learning, numerical simulation, and robotic synthesis. It generates, tests, and optimizes block by block.

At the heart of the system are ML potentials, trained on quantum calculations, which track every ionic ‘car’ on timescales inaccessible to ordinary quantum mechanics. They watch how mixing entropy transforms ordered rows into a labyrinth, where chaos gives birth to high-speed overpasses — correlated ion hops. Impedance spectroscopy acts as traffic cameras, recording real flow through electromagnetic responses. And Boltzmann’s law turns the steepness of energy hills into the probability that an ion will dare to climb. Even hydride electrolytes with their unusual anion mobility and plasma spraying of thin films naturally weave into this traffic pattern. Schrödinger’s wave mechanics refines the density functional to a shine — ensuring the city map is accurate down to the last atomic alley.

The practical payoff is impressive: the DigBat city archive already stores traffic maps for 3,736 electrolytes, and screening 32 million designs pinpoints 583 promising master plans. NaₓLi₃₋ₓYCl₆ halides deliver conductivity up to 10⁻³ S/cm at 100°C — almost liquid-like. A robotic builder stamps out over 100 blocks a day, instantly reporting successes and failures to the AI planner. Mistakes are put to use: each dead-end interchange teaches the AI urbanist not to repeat miscalculations.

The outlook is both enthralling and slightly unsettling. Autonomous discovery frees people from the drudgery of trial and error, but raises new questions. How can we unify fragmented data so the AI doesn’t confuse residential areas with industrial zones? Can the system predict not only bulk traffic but also accidents at interphase boundaries — the eternal problem of dendrites and jams from space-charge layers? Generative models are already eyeing unexplored architectures, and high-entropy mixtures promise compromises we haven’t yet imagined. Solid-state batteries are shedding their alchemical past and becoming an exact science: every ion trajectory is computed before the first sample goes into the furnace. But when every path is laid out by an algorithm, do we not lose that random spark that has ignited scientific revolutions for centuries?