Conclusion first

• I have tech ideas and prototypes in AI roleplay and training, and AI feedback on someone's social media profile. I've focused on training people to persuade via "advanced techniques" (not really, but people don't do them) of active listening and bridge building. Political campaigns are a likely target customer.
• I am not getting anywhere without a product manager, enthusiastic first customer, organization working on this problem who wants my tech, or a similar partner.

My background

• My inspiration is EA-like. I have the career capital (mid-career SWE) to take the risk, why shouldn't I? Civics and pro-democracy is the most compelling big problem to me, and consistent with EA, I'd like to try putting my career toward a problem I care about from first-principles like "make the world a better place."
• So right now I am in transition, having quit my previous job, identified as a startup founder (lol) and taken contract work in the AI/civics space (less prestigious, more "actual work.")
• My startup idea is deep canvass training. I've volunteered and coached in deep canvassing since 2020. This type of bridge-building persuasion with people is fun, deeply moral, and effective.
• Training has been a barrier to adoption, and training can be scaled via AI roleplays.

How things have gone

• I've not found early partners in deep canvassing. At some point, I will need a good partner, who can be a co-founder, org I contract for, or enthusiastic early customer who implicitly does some product managing as they work out their features and I work out my capabilities.
• I did work a contract pertaining to an AI in civics problem. That door was opened by my deep canvass demo. This is somewhat encouraging in that there is movement in the space but didn't turn to a longer term engagement.
• I got in touch with Higher Ground Labs at a weird time because I was already in a contract.
• YC networking: unpromising. Too many people are trying to make an unbiased news app, without grasping that people seek biased news deliberately.
• Right now I'm feeling between "focused" and "pessimistic." The "focused" thing to do is find a good first partner and pause on tech experimentation. Plan B is pessimism and go back to big tech.

Demand to be a persuader

• People really like persuading people of things. I think this fact is neglected. You can train people to persuade better, including on highly partisan topics, and they like doing it. However, people mostly just demonize the other side, and even sound like they've given up on the fact that persuasion is possible.
  • One AI tech option is a roleplay, which I've built for deep canvassing.
  • Another is a social media profile analyzer that points out some occasions you did get a bit more persuasive than other occasions.
• But if I steelman "this is a dumb idea," the first thing I notice is there are a large number of organizations (start here: https://www.turntoward.us/who-we-are) working on this. So the question is if this can be tech-enabled, in a way that it isn't yet, by working with the right partner.
• So yes, consistent with my problem statement above: good partner > more ideas.