About Cognograph

The Product

Cognograph is a spatial AI workflow canvas: a tool for people who think by arranging things. Instead of linear chat threads, you work on an infinite canvas where every conversation, terminal, note, and artifact is a node. Connect them with edges and context flows between them automatically.

Ask a question in one node, and it already knows what's in the nodes you've connected to it. Run a terminal command, and the output feeds into the next conversation. Generate an image, and it appears as an artifact you can connect to anything else.

The Builder

Cognograph is built by Stefan Kovalik, a full-stack designer and developer with 15 years of experience and a background in Cognitive Psychology from UC Santa Barbara.

Stefan's work sits at the intersection of how people think and how software should respond. His methodology, Perception-First Design™, applies processing fluency, decision architecture, and cognitive load theory to build interfaces that feel right before you consciously evaluate them.

Cognograph grew from a frustration with linear AI workflows. Thinking isn't linear. It's spatial, associative, and contextual. The canvas reflects that. Four provisional patents cover the spatial trigger mechanisms that make it work.

Open Source

Cognograph is open source under AGPL-3.0. You can inspect every line of code, audit the security model, and self-host if you prefer. The cloud service exists for convenience, not lock-in.

How Cognograph compares

vs Flowith

Flowith gives you a canvas of 40+ curated AI models. You work with what they bundle, on their cloud, on their subscription. Cognograph gives you a canvas where you bring your own keys. Claude, GPT, Gemini, or Ollama for zero-cost local inference. Per-chat provider switching. Your nodes, your models, your machine if you want it.

vs voicetree.io

Voicetree is a spatial graph of markdown notes and terminal agents with radius-based context retrieval. Cognograph is an orchestration layer across the same primitives with BFS graph traversal, 13 spatial trigger types, and four provisional patents on the trigger mechanisms. The canvas is the context engineering configuration. You engineer what context gets called, from where, at what depth.

vs Obsidian

Obsidian is a notes app with a graph view. The graph is visual. The AI can't read it. Cognograph makes the graph functional. Every connection is a context edge the AI traverses. Your thoughts become automated context engineering. If you don't need the AI collaborator, keep using Obsidian.

vs n8n

n8n is workflow automation. Trigger, action, action, done. Cognograph does this too (cron triggers, orchestrator nodes, autonomous agent execution), plus it holds state across sessions, plus conversations and terminals and artifacts live on the same canvas. n8n's functionality is one layer inside Cognograph. Most users don't need the workflow surface, but it's there when they do.

vs Weavy / Google Stitch

Weavy (now Figma Weave) is a node-based media pipeline. Google Stitch is a Gemini-backed UI design generator. Both are specialized vertical tools for specific outputs. Cognograph sits upstream as an orchestration layer. You can run a media pipeline inside it. You can generate UI inside it. But you can also see and modify the design wrapper, so you avoid the same slop output. One canvas covers image, video, code, conversations, terminals, and tasks.

Contact

General inquiries: [email protected]

Bug reports and feature requests: GitHub Issues