Yves Gugger
Yves Gugger is an individual designer, technologist and solo developer/creator whose principal current project is LeanCTX, an open-source context gateway for AI systems. LeanCTX sits between AI tools and the information they read to select relevant context, apply supported controls, and make usage and outcomes more traceable.
At a Glance
- Individual developers and AI power users
- Software engineering teams using coding agents
- Organizations with confidential or regulated information
- Enterprise teams operating AI workflows under shared policy
- +1 more
AI Tools by Yves Gugger
(1)
LeanCTX
AI Context Gateway for Developers
Discussions
No discussions yet
Be the first to start a discussion about Yves Gugger
Latest News
LeanCTX 3.11.4 bugfix release improved near-miss tool-call handling, Claude Code plugin wake-up behaviour and cockpit savings visualization.
LeanCTX 3.11.0 Community Engine release.
v3.10.2: More reliable reads and tool calls, including bounded retries for transient file locks and MCP roots-probe timeout handling.
v3.10.1: Packaging and runtime reliability improvements.
Products & Services
Open-source, Apache-licensed local context gateway/engine available through CLI, MCP, hooks and proxy paths. It selects and prepares relevant information, applies local safety and context controls, supports reference routing, usage accounting, cache, memory and recovery, and records evidence/receipts.
SDK for embedding supported LeanCTX context selection, controls and evidence into applications. Official documentation describes ContextSession, AgentContext and Agent Tools contracts, with Python, TypeScript and Go interfaces; commercial embedding, redistribution and OEM rights require separate licensing.
Optional hosted service adding Personal Cloud sync and backup plus a 1 GB hosted index to the local Community Engine.
Licensed organizational deployment offering identity, roles, access controls, policy, audit and approval controls, deployment and operating support under an agreement.
Market Position
LeanCTX positions itself as a local-first, model-agnostic context gateway rather than a standalone model or ordinary vector-search product. Its differentiation is the combination of context selection, permissions and safety controls, governed execution, memory/cache, measurable cost and outcome signals, and evidence/receipts. Relevant alternatives include native context/RAG layers, vector databases and search platforms, and agent gateways such as LangChain/LangGraph, LlamaIndex, Continue and similar coding-agent context tools; LeanCTX emphasizes local control, portability and operational governance across these workflows.
Leadership
Founders
Yves Gugger
Designer and technologist based in Zürich; trained originally as a mechanical engineer, with a master's degree in design. His portfolio describes prior work with Adnovum, Mobility, GenTwo, Oepfelbaum, SBB and Zühlke, including seven years of mission-critical UX at Adnovum and product leadership at GenTwo. GitHub identifies him as the creator behind yvgude and LeanCTX.
Founding Story
Gugger says he started building LeanCTX after repeatedly seeing AI systems read too much and waste context. The initial vision was a layer beneath different AI tools that would reduce repeated context, improve efficiency and transparency, and keep users in control of their data and tools. LeanCTX is developed through Thinkery, which describes itself as an independent product company; the project remains attributable to Gugger rather than being treated as a separate person or company named Yves Gugger.
Business Model
Revenue Model
Open-source Community Engine distributed under Apache 2.0; optional Pro subscription, paid Enterprise licensing/deployment, one-off onboarding services, and separately negotiated SDK/OEM embedding and redistribution agreements.
Pricing Tiers
Local Engine with sources, search, read modes, local safety, reference routing, budgets, cache, memory, receipts and recovery; no Thinkery account or private runtime required.
Adds Personal Cloud sync and backup and a 1 GB hosted index; 14-day free trial.
One agreed workflow in one test environment, including preparation, setup, team education and evaluation.
Organization identity, shared policy, reporting and deployment/support scope defined by written agreement.
Target Markets
- Individual developers and AI power users
- Software engineering teams using coding agents
- Organizations with confidential or regulated information
- Enterprise teams operating AI workflows under shared policy
- Product builders seeking embedded context controls and SDK/OEM rights
- AI-assisted software engineering and coding agents
- Company and enterprise knowledge workflows
- Multi-repository and external-provider context retrieval
- Legal, manufacturing, financial-services, healthcare and public-sector workflows
- Privacy-sensitive or data-sovereign local AI deployments
- Embedding governed context selection into third-party products