Home

Welcome to the Generative-Lives-Research-App wiki!

The Generative-Lives-Research-Assistant is conceived as a Socratic research tool powered by Gemini. It is designed for a specific use case, which is to assist me (Colin Greenstreet) in analyzing the impact of GenAI on professional training and hierarchies. It is a very specific use case of a potentially more generalized tool, which, if implemented, would encourage users to engage in Socratic dialogue and forensic analysis with a LLM, in this case Gemini.

I kicked off the development of the tool on Friday, November 21st 2025, just three days after the launch of Google's powerful Gemini 3 Pro Preview in the US and Europe. My intention was (and is) to test Gemini 3's broad conceptual, forensic, and coding capabilities. I am doing the development work in Google's AI Studio, using Code Assistant as my technical architect and coder. I have done some Red Hat testing using an instance of ChatGPT 5.1 in Thinking mode.

The initial brief to Gemini, which produced V1.0 of the model was:

"I want to build a research tool which will help me do the web research to support an article I am writing for my Substack Generative Lives. The theme of the article is the public statements and evidence of professions about their specific uses of generative-ai and how they are training new entrants at associate level, and different tiers of experienced professionals between entry level and partner. The research tool should be capable of supporting in depth research and should be Socratic. I want the tool to help me the user think conceptually about the problem, and to think where I the user (helped by Gemini) might find evidence."

After three days we are now at V4.4, which appears stable, and is ready for some human beta testing.

Colin Greenstreet will be discussing the concept, design, and ongoing development of the Research Assistant at the first meeting of the new ai-and-history collaboratory on Tuesday, December 9th 2025.

We have developed a User Manual and Beta Tester Protocol, which you can find in the GitHub repository associated with this wiki, or by clicking here.