How to Build an AI Second Brain for Product Work

TL;DR

Create a separate Claude or ChatGPT project for each workstream, then load it with relevant files, clear instructions, and ongoing conversation threads. These project-specific second brains retain context, reduce the burden of switching between initiatives, analyze customer feedback, improve work based on manager input, and support realistic product interview practice through voice mode.

Transcript

You've actually set up these independent, fully loaded brains in Claude and Chat GPT and then use those to hold the context for you and be more efficient with your job. So, show us what one of these brains actually looks like. Yeah, let's dive right into it. So, here's an example of a project. You can see here that I have a bunch of them. And in it... Read More

Key Insights

• Project-specific AI brains are separate Claude or ChatGPT workspaces containing files, instructions, and persistent conversation threads for individual initiatives. They reduce the mental burden of remembering every detail when a product manager moves between meetings, requirements, stakeholder requests, and multiple concurrent projects.
• The starting point for a useful AI project is reference material that defines the situation and desired outcome. Amir loads any available data, such as company kickoff decks, team kickoff decks, pull request references, and other materials that clarify what the project concerns.
• The value of persistent AI context is faster recovery after context switching. Instead of reconstructing an initiative from memory, a product manager can return to a project whose files and discussion history already capture its goals, background, open questions, and previous work.
• External customer conversations are a source of less company-directed evidence. Amir turned to Reddit because he wanted opinions about AI agents that were not tied to his employer's internal narratives, allowing him to examine what people independently discussed, expected, and wanted.
• Claude can guide a nontechnical user through building a data collection workflow in small, explicit steps. It helped Amir compare online sources, create a Reddit developer account, configure his environment, install packages, enter credentials, run commands, and generate the required scraper script.
• The Reddit research workflow produced 34,000 rows of conversations connected to Monday.com, AI, agents, and Amir's chosen search topics. This dataset gave his AI projects a substantial body of customer language that could be analyzed for patterns, priorities, expectations, and discussion points.
• The mindset that everything is text expands what can become AI context. PDFs, slide decks, project references, customer discussions, and other work artifacts can be added to project-specific systems so the model can reason from relevant information instead of relying on isolated prompts.
• Custom GPTs can target specific professional development needs. Amir uses them to improve skills based on feedback from his manager, while GPT voice mode provides realistic product interview practice, asks questions conversationally, and gives candid feedback about the quality of his responses.

Questions & Answers

Q: How do you build an AI second brain for product management?

Create a separate Claude or ChatGPT project for each major workstream. Add clear instructions, relevant files, and source material that explains the initiative, then keep related conversations inside that project. Amir's example grew from a few files to 20 files. The accumulated context lets the AI retain background, support ongoing analysis, and help the product manager resume work after frequent context switches.

Q: What files should a product manager add to an AI project?

Start with any material that helps define the situation, objectives, and existing work. Amir mentions company kickoff decks, team kickoff decks, pull request references, and any other data connected to the initiative. The purpose is to give the AI enough grounding to understand what is happening, what the product manager wants to achieve, and what the ongoing conversations are actually about.

Q: How does an AI second brain reduce context switching?

A project-specific AI workspace stores the files, instructions, and conversation history for one initiative. When the product manager returns after meetings or work on another project, the relevant context is already assembled. This reduces the need to remember every thread personally and makes it easier to answer stakeholder requests, continue analysis, and recover the reasoning behind earlier decisions or questions.

Q: Why did Amir Klein use Reddit for customer research?

Amir wanted a voice of reason that was not connected to his company or shaped by a specific internal narrative about AI agents. Reddit contained external conversations where people discussed their interests, expectations, and needs independently. He chose it after exploring other online sources because LinkedIn and Twitter had more limited APIs, and some available options required payment.

Q: How can Claude help build a Reddit scraper without coding expertise?

Claude can break the task into detailed setup and implementation steps. For Amir, it explained how to create a Reddit developer account, open the terminal, install Brew and required packages, configure the environment, and run commands. It then wrote a script that used his Reddit client ID, client secret, and username to collect conversations matching selected subjects.

Q: What did the Reddit scraper collect for Amir's research?

The completed workflow generated a file with 34,000 rows of conversations. The scraper targeted discussions connected to Monday.com, AI, agents, and other topics Amir selected. He had envisioned obtaining thousands of relevant conversations, and Claude helped turn that goal into an automated process whose output could be added to his AI workflow for customer feedback analysis.

Q: How can customer conversations improve product decisions?

Customer conversations provide direct evidence about what people discuss, expect, and want. Amir sought this evidence because AI agents attracted many internal opinions and ambitious visions, while the underlying customer demand remained uncertain. Collecting external discussions gave him material that AI could analyze for patterns, priorities, and key discussion points, supporting product reasoning with voices beyond company stakeholders.

Q: How can AI help improve product management skills and interview performance?

A custom GPT can be configured around specific improvement areas identified through manager feedback, giving the product manager a recurring system for reviewing and strengthening work. For interview preparation, GPT voice mode can conduct a realistic mock product interview, respond conversationally to answers, and provide candid feedback. These workflows turn professional feedback and practice sessions into repeatable, context-aware learning tools.

Summary & Key Takeaways

• Amir Klein manages product work by creating separate, context-rich projects in Claude and ChatGPT. Each project contains instructions, reference files, and continuing threads related to one initiative. This structure gives him an external memory for complex workstreams, helping him respond efficiently when meetings, stakeholder requests, and competing priorities force frequent context switching.

• He used Claude to turn an idea for customer research into a working Reddit scraper. Claude compared possible sources, provided detailed setup instructions, identified required packages, and wrote a script using Reddit credentials. Following the process produced a file containing 34,000 rows of conversations related to Monday.com, AI, agents, and selected topics.

• The broader workflow treats available information as text that can become reusable AI context, including PDFs, slide decks, customer conversations, and internal references. Amir also creates custom GPTs to improve skills highlighted by manager feedback and uses voice mode for mock product interviews that challenge his answers and provide candid response feedback.