Viktor.comAn AI coworker that actually does the work
Promoted
engrams
Feed knowledge into a brain that thinks for itself
14 followers
Feed knowledge into a brain that thinks for itself
14 followers
An engram is a living knowledge base that organizes itself. Feed it sources - URLs, papers, notes - or sync from GitHub, Notion, and more. It compiles everything into a knowledge graph and wiki, maintained entirely by AI. Every question you ask makes it smarter. Background agents tend to it while you sleep, refreshing information and discovering knowledge gaps. Obsidian and Notion make you the librarian. Engrams makes you the researcher.
why can't I just make my own AI Agent for this?
@rakesh_jhunjuhnwala you totally can and thats basically how this started. the question is whether you want to spend weeks building and maintaining the compilation pipeline, the indexing, the graph, the UI, the background agents, or just use something that already does it. same reason people use notion instead of building their own wiki in markdown
Interesting concept, but what architecture are you using for memory?
@sd393 No vector DB or RAG in the traditional sense. The LLM compiles sources into actual markdown articles incrementally, one source at a time against a lightweight index of what already exists. By query time you're reading structured articles, not retrieving chunks. Happy to go deeper on the compilation loop if you're curious.
how is this different from just dumping papers into chatgpt and asking it to summarize everything
@gama_adorama engrams compiles everything into articles that persist, link to each other, and get richer every time you add something or ask a question. its the accumulation that matters not any single response