Sebastian Rodriguez and Michael Cohen led a practical deep-dive into the operational reality of AI in modern campaigning, identifying a critical asymmetry in the current political landscape. While AI has become "ingrained" in the tech stack of modern races, the speakers noted a significant reluctance among progressive and pro-European campaigns to deploy these tools. This hesitation has created a vacuum that far-right parties, such as the AfD, have filled by dominating the "volume" of AI-generated content. Cohen observed that in the U.S., where national regulation has "completely failed," the fear of "persuasion prompting"—a technique proven to be incrementally more powerful than human messaging—has paralyzed mainstream parties, leaving the field open for unregulated experimentation by their opponents.

To bridge this gap, the session pivoted from theory to a hands-on "Launch a Campaign with 10 Prompts" exercise, aiming to demystify the technology. Rodriguez argued that the most effective and ethical entry point for hesitant campaigns is satire and humor. Because voters on social media seek "fun content" rather than "high-minded speeches," campaigns can leverage AI to create comical content that is clearly distinct from malicious manipulation. The speakers demonstrated this by using advanced reasoning models (like OpenAI's o1) to instantly structure a mock "Pirate Party" campaign based on "pagan utilitarianism," generating high-quality visuals—such as caregivers drowning in paperwork—that reflected specific local political obsessions while remaining obviously satirical.

The session concluded with a forward-looking analysis of "agentic AI" and the necessity of "Deep Research" models. The consensus was that the future of political tech lies in parties standing up their own "independent LLMs" to bypass the guardrails and biases of third-party corporate platforms. However, the speakers emphasized that the "human component" remains non-negotiable for auditing results to avoid hallucinations. The bottom-line actionable advice was to move toward systems that prioritize transparency and audit trails, using AI for rapid experimentation while adhering to a simple rule: "make it so comical that it is clear that... you're not trying to manipulate someone."