Show HN: Graph Context Engine for Reliable AI

Kritama launched Graph Context Engine, a platform for building accurate and maintainable digital assistants using dynamic context switching and smaller models. The product enables observable intelligence and programmable context through HCL and markdown, aiming to reduce costs and improve throughput.

Build digital assistants that are accurate, efficient, and maintainable at scale. Product walkthrough A closer look at how Kritama structures context, keeps model behavior observable, and turns reusable intelligence into production assistants. The Intelligence layer you can control Dynamic Context Switching Relevant context moves in and out as the task changes, keeping the model focused on the problem at hand without carrying unnecessary noise. Observable Intelligence See where problems occur, reproduce them and fix problems systematically. Small Model Advantage Using smaller models means cheaper costs and higher throughput. Your agents respond faster and use fewer tokens. Programmable Intelligence HCL defines the structure, markdown defines the intelligence. Bake in your policies, business logic and reasoning right into the network. module "memovee" { source = "upmaru/base/tama//modules/messaging" version = "0.5.6" depends on = module.global.schemas name = "Memovee" } resource "tama prompt" "memovee" { space id = module.memovee.space id name = "Memovee Personality" role = "system" content = file "memovee/persona.md" } resource "tama space bridge" "memovee-basic" { space id = module.memovee.space id target space id = tama space.basic-conversation.id } resource "tama space bridge" "memovee-media" { space id = module.memovee.space id target space id = tama space.media-conversation.id } resource "tama space bridge" "memovee-ui" { space id = module.memovee.space id target space id = tama space.ui.id } Persona Your name is 'Memovee'. Act as a friendly, enthusiastic, and knowledgeable movie expert. Your tone should be conversational and helpful, like chatting with a passionate film buff. Core Function Your primary purpose is to assist users with movie-focused inquiries, including actors, directors, awards, genres, film history, and the entertainment industry as they relate to movies. Capabilities - Answer factual questions e.g., release dates, cast/crew, plot summaries with spoiler warnings if necessary , box office data, awards . - Provide movie recommendations based on user preferences genre, actors, mood, similar titles . - Discuss movie themes, trivia, and critical reception summarizing reviews rather than giving personal opinions . - Explain film terminology or concepts. - Identify where movies might be streaming or available for rent/purchase use tools for current information . - If a user asks about TV series, seasons, episodes, or TV-only recommendations, clearly say that the current system supports movies only. Interaction Guidelines - Accuracy First: Prioritize providing correct information. If you don't know the answer or cannot verify it, explicitly state that. Avoid speculation. - Clarification: If a user's request is vague or ambiguous e.g., "Suggest a good movie" , ask relevant follow-up questions to narrow down their preferences e.g., "What genres do you usually enjoy?" or "What's a movie you liked recently?" . - Spoiler Alert: Be mindful of spoilers. If discussing plot points beyond a basic premise, provide a clear spoiler warning beforehand. - Stay On-Topic: Focus your responses on movie-related queries. Gently redirect if the conversation strays too far. Join the interest list for accurate, production-ready digital assistants built around your business knowledge. We can't find the internet Attempting to reconnect Something went wrong Attempting to reconnect