This capstone session turns the digital twin from a measurement instrument into a design tool. The same architecture that reads opinion can be run as a closed loop: propose a message, have a panel of LLM twins evaluate it, and search for the version that resonates best with a target audience. We work the idea on two grounds. First political campaigns — generating and optimising party platforms against a twin panel. Then the courtroom, where the same machinery is re-pointed at legal argument: twins of jurors and judges, a synthetic mock trial, and verdict probabilities read as the case is re-argued with different opening framing, evidence order and narrative emphasis.

Participants work hands-on with a pre-built twin panel, putting a set of candidate messages to it, aggregating the evaluations and identifying which performs best — one pass of the optimisation loop, done by hand. We then demonstrate the full closed loop running over a much larger candidate space, and explain how the search works. Throughout we keep validity and ethics in view: how to validate synthetic jurors against the experimental mock-trial literature, how the contamination of well-known cases threatens prediction, and why these are tools for stress-testing arguments before they are made, never for replacing adjudication.

Aimed at researchers interested in message testing, persuasion and strategy — across political communication, legal research and computational social science. Participants should be comfortable running R notebooks; no legal or machine-learning background required. The morning modules are helpful but not required.