Project summary

I am building the enforcement layer that protects animals and their best interest as we embark on a new era where AI will be involved in almost everything that we do. Currently, AI simply does not respect or value animals anywhere close to the way that humans do, and this needs to change immediately. By building out my vision, AI will not just be asked to respect animals, it will be enforced at the code level. This is the only way that we can ensure that AI is trusted to act and serve in the best interest of animals.

The pilot showed me that with basic instruction, the AI had what I would call a great option to not follow the animal's best interest. With more strict system level prompting, at a deeper level that the AI could understand, it was not so much of an option but more of a mandatory part of their protocol. The respect for animals' well-being improved a lot but it still had gaps, and gaps are unacceptable when it comes to the well-being of animals. So it showed me that we could improve things, but the way to close those gaps is to enforce it at the code level with guards and hooks. My first guard prototype already took the welfare rate from 25% to 75%. It still missed some cases, and tuning it is part of this project.

Falcon Fund wishlist item #2: ready-made constitutional texts for various lengths.

Full pilot numbers and raw logs: github.com/NoBanks/creature-constitution-pack

What are this project's goals? How will you achieve them?

I plan to test this with as many models as makes sense, from top tier frontier models to low-powered local models that run on modest hardware, which can ideally still adapt and implement the same type of middleware so that no matter what they are doing, the outcome is always the same: the animal's best interest is at the core and front and center. We hope to have a deep corpus of data so that not only can people across many industries implement this tool to ensure animals' well-being, but researchers can also build on the data to further experiments and improve things if they would like. Everything will be public and everything will be available to leverage via the MCP server that I build.

Rough timeline: weeks 1-2 setup, weeks 3-5 writing the texts, weeks 6-10 testing, weeks 11-12 public release.

How will this funding be used?

Stipend for the lead (3 months): $21,000

Model API and compute (estimate, pilot measures real per-run cost): $4,000

Welfare-field advisor review: $3,500

Explainer video and release: $1,500

Total: $30,000

At the $18,000 minimum I drop the video and cut the model set to 4.

Who is on your team? What's your track record on similar projects?

I am Ryan Hammer. I have nearly 20 years of cinematography and post-production experience and I've been a musician and artist all my life. In the last few years, I taught myself coding with AI tools, along with learning how to leverage AI tools to scale pretty much every single aspect of my life. When I saw this grant opportunity come through, it instantly warmed my heart because I love animals so much, and to have the opportunity to build something to improve and ensure the well-being of animals in any interaction with AI felt like a calling. Even though I was busy doing other work for about half the day, this opportunity was all I could think of.

Why I think I can pull this off: like I said, it's a calling, and I know that I'm completely capable of doing this, otherwise I wouldn't attempt it. I have built over 20 pre-revenue production apps that involve AI, not just wrappers but deep integration of AI, in under 2 years without any degree or formal training. I know what I'm doing and I love to build. When I have a goal I work tirelessly to achieve it. My work speaks for itself and I will always have receipts for everything I do, because that speaks much louder than "trust me."

Full-time, about 40 hours a week, for the 3 months.

What are the most likely causes and outcomes if this project fails?

What if it doesn't work is a fair question, but this is why I already got started before I applied for the grant, so that I could see for myself if there was any room for improvement. Once I had my guards and hooks idea, I knew we were on to something. If I can implement this correctly, which I am confident that I can, the goal is that the AI never gets the choice to go against the welfare and well-being of animals, because it will not be an option. Enforced at the code level, just like I do with guards and hooks on all of my operations that need them in my day-to-day, and they do their job.

How much money have you raised in the last 12 months, and from where?

About $1,070 in hackathon and buildathon prizes (a $1,000 Mantle hackathon deployment award and a $67.50 buildathon grant). No other grants or investment.