6/ OpenAI and Thrive Holdings are already working closely together.
One example: the tax agents we recently built with Codex.
http://openai.com/index/building-self-improving-tax-agents-with-codex
5/ You can’t answer that from the outside.
You need real companies, real workflows, real users, and real constraints.
Users are excited about OpenAI and Thrive Holdings building self-improving tax agents with Codex because it advances their Long Humans thesis on future AI work.
7/ OpenAI’s equity stake in Holdings also makes the alignment unusually strong.
Deployment informs product. Product informs research. Research improves models. The loop gets tighter the deeper the work gets.
6/ OpenAI and Thrive Holdings are already working closely together.
One example: the tax agents we recently built with Codex.
http://openai.com/index/building-self-improving-tax-agents-with-codex
8/ My role is to help with that loop: technical direction, real-world measurement, and turning what we learn in deployment into faster product and model improvements.
9/ Excited for what we build next.
Thrive Holding’s Long Humans post goes deeper on the thesis behind the company:
http://www.thriveholdings.com/long-humans
6/ OpenAI and Thrive Holdings are already working closely together.
One example: the tax agents we recently built with Codex.
http://openai.com/index/building-self-improving-tax-agents-with-codex
5/ You can’t answer that from the outside.
You need real companies, real workflows, real users, and real constraints.