Kenny Vaneetvelde
Kenny Vaneetvelde is an independent software engineer, AI practitioner, open-source developer, and writer working under the Eigenwise name. He helps companies get practical value from AI through hands-on workshops, focused implementation sprints, advisory work, and open-source tools that emphasize explicit control and production reliability.
At a Glance
- Companies and engineering teams adopting AI
- Software teams using Claude Code and MCP
- Developers building production agentic applications
- Organizations seeking AI workflow training, custom integrations, and architecture guidance
AI Tools by Kenny Vaneetvelde
(1)Atomic Agents framework
Modular Python AI Agent Framework
Discussions
No discussions yet
Be the first to start a discussion about Kenny Vaneetvelde
Latest News
Published ‘The AI Demo Worked. Now Show Me the Cleanup.’, arguing that demos and token volume do not establish production readiness.
Published ‘Stop Teaching Your AI Agent the Wrong Lessons’ about memory, instruction placement, and evaluating whether saved lessons help.
Published a hands-on Atomic Agents and Phoenix tracing guide, showing how to inspect planner, calculator, and answer-writer traces.
Published a guide to the Eigenwise Toolshed codebase-mapper plugin for giving Claude Code a compact project index and source pointers.
Products & Services
A lightweight, modular Python framework for building AI agents and agentic pipelines. It uses Instructor and Pydantic for structured inputs and outputs, keeps control flow in application code, and avoids hidden LLM calls and opaque abstractions.
A MIT-licensed marketplace of Claude Code plugins for recurring engineering work, including codebase mapping, scoped rules, ticketed parallel work, model access, usage monitoring, workspace setup, observability, and quality gates.
A TypeScript SDK, WebSocket protocol, and MCP gateway that exposes typed live web-app actions to MCP-compatible agents, with confirmation, elicitation, progress, sampling, resources, and session capabilities.
A one-day remote, hands-on AI workshop using a customer's tools, codebase, and real use cases; includes Claude Code, skills, MCP fundamentals, and a written recap.
Market Position
Kenny positions Atomic Agents and his consulting against opaque, abstraction-heavy agent stacks such as LangChain, CrewAI, AutoGen, and PydanticAI. The differentiators he emphasizes are atomic components, explicit application-owned control flow, structured schemas, provider flexibility, inspectability, and production maintainability rather than maximum abstraction.
Founding Story
Kenny says he created Atomic Agents out of frustration with popular agent frameworks that hid important behavior behind abstractions and with systems that broke in production. His broader Eigenwise work applies the same developer-first, inspectable approach to helping teams adopt AI in their real workflows.
Business Model
Revenue Model
Consulting and advisory fees from workshops, implementation sprints, and ongoing advisory engagements. Open-source software is free to use; maintenance can optionally be supported through GitHub Sponsors or Ko-fi, while customers separately pay their own model-provider costs.
Pricing Tiers
One day, remote.
Six working days spread over about four weeks.
Ongoing, about two days per month.
Target Markets
- Companies and engineering teams adopting AI
- Software teams using Claude Code and MCP
- Developers building production agentic applications
- Organizations seeking AI workflow training, custom integrations, and architecture guidance
- Building maintainable AI agents and multi-agent pipelines
- RAG chatbots, web-search agents, deep-research agents, multimodal workflows, and YouTube summarization
- Tracing and debugging agent, model, and tool calls in production
- Adding AI to engineering workflows and Claude Code projects
- Exposing actions in live web applications to MCP-compatible agents
- Team AI training, workflow audits, internal-tool integrations, and AI architecture advisory
- Contractify: Kenny replaced a LangChain foundation with Atomic Agents and built Ada, a production data-extraction agent.
- Henkel: Kenny designed and delivered an MCP workshop for a team.