Release notes · Research · Essays

Latest.
Everything we publish.

Release notes for every Kaptain version, the benchmarks behind our numbers, and the research on token economics, delegation and agent systems. Newest first.

Release notes Kaptain release notesWhat changed in each release. Read · Guide Choose your partsWhy AI work gets better when you can swap the model, tools and helpers on their own, and how that keeps you free of lock-in. Read · Guide Krew and MCP: how models get their toolsWhat a tool call is, why the open Model Context Protocol matters, and how Krew gives every model the same tools behind one approval gate. Read · Guide What is a kapsule?Agent harnesses from first principles, how a kapsule packages one, and how several kapsules work together. Read · Essay How Kaptain reduces unnecessary token spendDelegates, local options, continuity and granular model control. Read · Capability statement Kaptain capability statementToken-cost pressure, the KodeGraph benchmark evidence, the launch model and the expected impact. Read · Essay When AI usage costs more than the work it supportsA careful look at when sustained AI operating cost can rival the workflow it supports. Read · Benchmark KodeGraph helps Kaptain find the right code with fewer tokensRetrieval quality, token cost and system load of Kaptain's built-in code graph, measured against a standard code-graph tool. Read · Whitepaper The Agent Ecosystem OSVersion 1.1: a practical architecture for connected agents, model routing, token offloading and measurable workflow control. Read · Essay Token costs are becoming an operating issueWhy recurring agentic workloads make model routing an architectural concern. Read

Looking for how to install, set up or run Kaptain? Everything lives on one page: Kaptain Docs.