Launched this week
OttO
Graph-Based Self-Updating MCP Knowledge Base.
8 followers
Graph-Based Self-Updating MCP Knowledge Base.
8 followers
OttO is a graph-based semantic knowledge database that runs entirely on your local device. It connects to other AI tools as an MCP server and continuously updates its knowledge graph based on the insights and inferences those tools generate. Instead of spending extra tokens on tasks that have already been completed, OttO reuses existing knowledge and expands it with the information you provide, allowing its knowledge graph to grow and improve over time.
How does Otto handle conflicts when multiple AI tools feed it different or contradictory inferences about the same entity at the same time?
@linavzbw Thank you for the question. Different inferences about the same topic are actually beneficial for OttO because they expand its knowledge and coverage. When a new inference is added, OttO can use it to answer future questions that are related to that topic. As long as the coverage expands, the specific way it expands is not important. The matching is not performed as question–answer pairs; it is based solely on the inferences by ai tools and the information provided by the user.
How does OttO handle conflicts when two connected AI tools generate contradictory inferences about the same entity in the graph?
@ravzaerbeyler OttO also sends the semantically related nodes connected to the content. When new content is added and the inference differs, the connected knowledge graph nodes are updated accordingly. As a result, different inferences do not mislead the model. Instead, they expand the context, enabling the model to handle alternative interpretations more effectively and comprehensively.
Tried OttO with my usual MCP setup and the local knowledge graph actually picking up inferences between sessions was a nice surprise. feels like my tools finally remember things without me feeding them context again.
@hasboyac54710 I'm glad you're enjoying the app. I'm actively working on new features, and seeing kind feedback like this genuinely makes me happy.