I am building Reznull, a product that uses AI personas to run fast, simulated message tests for B2B SaaS marketers and messaging agencies. Given several message variants, a target buyer, and a desired action, Reznull estimates which variant is most likely to make that buyer act.
Reznull evaluates the variants using AI personas built from public evidence and returns a recommended winner with a confidence level, or no winner when the evidence is too weak.
Software can now be built in days, while reliable user feedback still takes weeks. Many B2B SaaS pages and narrow buyer segments do not have enough traffic for a useful A/B test. Human panels can provide evidence, but they are expensive and require setup and analysis for every study. Reznull aims to let teams evaluate more ideas before deciding which messages to use or validate with real users.
I have spent the past six years building production ML products and experimentation systems. You can read about that work, along with my independent projects, talks, and writing, on my personal website.
If you invest in early companies or are interested in helping build Reznull, I would be happy to talk.