# Yamil Velez

- **Event:** [Built with Opus 4.7: a Claude Code hackathon](https://cerebralvalley.ai/e/built-with-4-7-hackathon)
- **When:** Apr 21 at 12:00 PM – Apr 27 at 2:00 AM (EDT)
- **Where:** Online
- **Team:** [Yamil Velez](https://cerebralvalley.ai/u/yrvelez)
- **Demo video:** https://youtu.be/NgbObhKR0NA
- **Gallery:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/169

Nationscope AI is a rapid message-testing platform that compresses what traditionally takes campaigns and advocacy groups weeks — testing ad/messaging variants — into minutes at near-zero cost. It targets political campaigns, advocacy orgs, public-health agencies, and newsrooms. The pipeline (ideate → screen → field → export):

Lab Mode — Opus 4.7 strategist. You give a brief, the system uses Opus to cluster Haiku-generated persuasion angles via semantic similarity, then writes diverse, non-redundant variants. A "Max Creativity" toggle pushes for further-out theories.

Focus Group — 8 Haiku avatars. Demographically-weighted synthetic personas react to each variant in parallel, streaming sentiment in real time. Haiku voices each persona; Opus synthesizes themes, objections, and survivor picks.

Poll Mode — silicon experiments. Randomized A/B/C trials over 500K Nationscape personas (UCLA + Democracy Fund data: real demographics, Big Five traits, Schwartz values, baseline policy stances).Placebo arms isolate treatment effects.

MRP adjustment. Multilevel regression + poststratification with ACS weights projects synthetic-sample estimates onto the actual U.S. population (51 fitted policy targets cached at boot).

Qualtrics export. One-click handoff of the same arms, outcome measure, placebo, and randomization to a real human survey — so the synthetic pre-test maps cleanly to a live field study.

---

Markdown version of https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/169. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt