The Brains: Turning Founders into Queryable MCP Servers

When You Want to Know How Pieter Levels Prices

When pricing a product, you might want to reference Pieter Levels' thinking; when working on growth, you might want to see how Ryan Hoover cold-started Product Hunt back in the day. This information is scattered across podcasts, blogs, interviews, and tweets, making search-and-scroll extremely inefficient. The Brains, ranked #8 on Product Hunt, offers a new solution: turn founders, CEOs, and investors into MCP servers that plug directly into Claude, Cursor, ChatGPT, or Codex for Q&A.

Its tagline is blunt: "Founders and investors as MCP servers, every answer cited." It currently hosts 37 "brains," each with its own URL, plus a unified entry point. The project is free and open source (MIT license), and explicitly states it is unofficial with no affiliation to the individuals themselves.

Product Mechanism: From Public Content to a Searchable Knowledge Base

The Brains works by scraping and organizing the target person's public talks, articles, interviews, and social media posts into a searchable knowledge base, then exposing it to AI clients via the MCP (Model Context Protocol). When a user asks a question in Claude or Cursor, the AI calls the corresponding "brain" to retrieve relevant snippets and returns an answer with source links.

The key difference from generic LLM Q&A lies in the word "citation": the model doesn't fabricate a founder's views from training memory, but extracts from that person's actual public corpus and attributes the source. For scenarios requiring firsthand quotes—writing articles, competitive research, preparing investment memos—this is quite practical.

MCP is an open protocol proposed by Anthropic that has gradually become one of the de facto standards for connecting AI tools to external data sources over the past year. The Brains packaging a "personal opinion library" as an MCP server is a rather interesting application direction in this ecosystem: not connecting to a database or API, but connecting to "someone's public thinking."

MCP (Model Context Protocol) is an open protocol proposed by Anthropic in late 2024, aimed at standardizing how AI applications connect to external data sources and tools. It uses a client-server architecture: AI clients (such as Claude Desktop, Cursor) act as hosts, obtaining context through resources, tools, and prompts exposed by MCP servers. Unlike traditional API integration, MCP lets developers implement a server once and have it called by multiple compatible AI clients, reducing repeated adaptation costs. Over the past year, MCP has gained support from a large number of tools and platforms, gradually becoming one of the de facto standards for connecting AI tools to external data. The Brains leverages this protocol to encapsulate personal opinion libraries as MCP servers, enabling AI to retrieve and cite on demand.

RAG (Retrieval-Augmented Generation) is likely the core technical paradigm behind The Brains. Its basic flow: first split the corpus into chunks and vectorize them into a database; when a user asks a question, the system retrieves the most relevant chunks and passes them as context to the LLM to generate an answer. RAG's key advantage is that the model doesn't need to memorize all knowledge in its parameters but extracts information from an external knowledge base in real time, reducing hallucinations and supporting citation traceability. The Brains' emphasis that "every answer is cited" is a typical RAG characteristic—the model answers based on retrieved original snippets and marks source links. However, RAG's effectiveness depends heavily on corpus quality, chunking strategy, and retrieval algorithms, which explains why the knowledge base's boundaries are limited by public corpus coverage.

Use Cases: Who Would Use It

Product and growth practitioners can query specific questions, such as "How does Pieter Levels price?" "How does Paul Graham talk about fundraising?" "How did Ryan Hoover launch Product Hunt?" Compared to generic "startup advice," this retrieval targeting specific individuals is closer to hands-on practice.

Content creators and researchers who need to cite firsthand opinions save verification costs with linked citations and reduce the risk of misattribution.

Investors and entrepreneurs can quickly understand an investor's public stance and focus areas when doing due diligence or preparing for a conversation.

However, it's important to recognize that the knowledge base's boundaries depend entirely on public corpus coverage. A person's recent views, expressions in non-public settings, or content outside the corpus scraping time window won't appear in answers. And a scale of "37 brains" means coverage remains limited—more curated than comprehensive.

Open Source and Ecosystem: Possibilities Under the MIT License

The project is open source under the MIT license, meaning others can reuse this approach to build brains for the people they care about, or integrate existing brains into their workflows. 37 brains, one URL per brain, one unified entry point—this structure naturally supports expansion.

From a product perspective, The Brains is more a reference implementation of a "protocol-layer application" than a product chasing mass users. Its value lies in validating an idea: in the MCP ecosystem, vertical, cited knowledge sources can become capability plugins for AI clients.

Risks exist too. Being an unofficial project means compliance boundaries around corpus usage need careful handling; the timeliness and context-dependence of personal views also require users to retain original links and judge for themselves when citing. The MCP ecosystem itself is still rapidly evolving, and the stability of protocol and client support is also a variable.

For developers, what's more noteworthy is the pattern it demonstrates: structuring public corpora, attaching citations, and connecting to AI tools via a standard protocol. This pattern can be replicated across many vertical domains—industry experts, academic researchers, even internal technical leads at enterprises.

Quick Start

If you're already using Claude or Cursor, you can find the project address on its Product Hunt page, get a single brain's URL or the unified entry point, and configure it as an MCP server in your client to start asking questions. Since it's open source, you can also inspect the source code to understand how corpus processing and retrieval are implemented.

When asking questions, it's recommended to include specific scenarios and names, such as "What is X's view on Y?", which is more likely to yield valid, sourced answers than broad questions.