The following is an idea that has been in the back of my mind for a long time. However, from a technical perspective it is now markedly easier to engineer with the appearance of GPT-4 and similar LLM APIs. If you've been thinking about something like this, please reach out! So much of the difficulty of achieving change in the world lies in issues of fragmentation – finding out who already had an idea, who wants to buy a product, who would want to work on something, who has experience/knowledge about a particular topic, etc. Currently, the best and only way to do this is via direct transactions – talking to others, writing emails, one-to-many communication (eg. newsletters, social media). The proposed idea is a system of interconnected note taking tools where you can get to know information from other people's notes in an anonymized and double opt-in way. The proposed tool introduces the concept of "conditional transactions." Status quo in digital communication • Alice directly contacts Bob with a message (text, email, calls). Alice needs to know the identity of Bob in order to contact him, and needs to pre-draft her message. • Alice broadcasts a message to a pre-defined group (chat groups, social media, newsletters). Alice needs to draft a general message aimed at a group (likely people she's friends with or have interacted with in the past). Proposed flow • Instead of picking a person to contact first and then composing a message, Alice broadcasts a query aimed at an idea – which is then backfilled into users that expressed that idea in the past. These users become the recipients in an anonymized, double opt-in manner. Example 1 • Alice has a note: "I'm considering opening an account at Schwab. I'd like to learn more from ideally someone who has used Schwab extensively and can compare it to other brokerages" • The tool then searches for other users with the query above. It selects user Bob with notes from 2018 onward related to Schwab: "set up an account at Schwab to put my savings in", "one year later, switching away from Schwab [...]" • Alice gets prompted: "A user was found with notes relevant to your query about experience with Schwab. Do you want to:" • "Request note access (originals)" – requests original snippets from notes • "Request note access (anonymized summary)" – requests AI-summarized snippets in a non personally identifiable manner • "Request identity" – request Bob's identity Example 2 • Alice has a note: "I'm working on Kubeprem, a company that offers managed Kubernetes systems for on-premise hosting of Kubernetes." • Bob enters a query: "I'm looking for a managed Kubernetes service for our server park" • Alice gets prompted: "Results found of notes relevant to your search about on-premise Kubernetes tools:" • "Request note access (original)" • "Request note access (anonymized summary)" • "Request identity" Alice can then request to learn more, and if approved from Bob, they can reveal their identity to each other. Considerations • The tool would need read access to apple notes, obsidian and other common note tools. A custom app could be built on top of these that exposes a minimal UI for user interactions. • A critical requirement to this tool working is people actually entering the notes. This would likely make it a niche tool that only a subset of thoughtful and disciplined people use, as the average consumer is unlikely to consistently and thoughfully journal. • This tool solves the issue of interconnectedness and fragmentation, allowing people to conditionally contact each other and provide a platform for people who want to genuinely sell and "be sold." However, users would still need to build trust on their own with other individuals they connect with.