Giving Claude Code and Codex one searchable memory on your own machine
Speaker: Henrik Ibsen, CEO of E-HealthBrains and Creator of NOBLIVION
Duration: 60 Minutes | Live Online | Free
About the Workshop
AI coding agents like Claude Code and Codex can write code, debug applications, and work across complex repositories. However, one of their biggest limitations becomes apparent when working on larger projects over extended periods: they don't reliably remember everything they've learned.
Important architectural decisions get lost between sessions. Lessons from previous mistakes aren't always retrieved when needed. And when switching between Claude Code and Codex, developers often have to explain the same project context all over again.
What if both coding agents could access the same persistent, searchable memory, stored entirely on your own machine?
In this free technical webinar, Henrik Ibsen, CEO of E-HealthBrains and creator of NOBLIVION, will explore how to build more reliable coding workflows using shared, persistent memory.
Drawing on his experience developing Phoenix Daemon, a larger on-premises agentic system, Henrik extracted NOBLIVION, an open-source project that provides Claude Code and Codex with access to a shared local memory store.
We'll explore why simply storing more information isn't enough, how relevant memories can be retrieved at the right moment, and how previous mistakes can become reusable lessons that influence future agent behavior.
What You'll Learn
Why coding agents forget: Understanding the limitations of session-based context, memory files, and conventional approaches to maintaining long-term project knowledge.
Building shared memory across coding agents: How Claude Code and Codex can access the same locally stored project notes, decisions, rules, and lessons without maintaining separate knowledge bases.
Retrieving the right memory at the right time: How searchable memory, semantic retrieval, and context injection help agents find relevant information without loading an entire history into the context window.
Learning from previous mistakes: How errors and developer corrections can become reusable lessons, and how guard mechanisms can help prevent known mistakes from being repeated.
Local-first memory architecture: How NOBLIVION uses a local memory store, integrations with coding agents, and memory usage tracking to provide more persistent and controlled agent workflows.
We'll also explore NOBLIVION's architecture and practical usage, including how the same memory can be accessed from Claude Code and Codex, and how relevant lessons can be surfaced during development.
About the Speaker
Henrik Ibsen is the CEO of E-HealthBrains and the developer behind Phoenix Daemon and NOBLIVION.
His work focuses on building deterministic, on-premises agentic systems with persistent organizational memory, controlled actions, and mechanisms for improving agent reliability over time.
While developing Phoenix Daemon, Henrik encountered limitations in how coding agents retained and applied lessons from previous sessions. This led him to develop NOBLIVION, a standalone open-source memory system designed to give Claude Code and Codex access to shared, searchable project knowledge.
Who Should Attend?
This workshop is designed for software engineers, AI engineers, developers using Claude Code or Codex, and anyone interested in persistent memory, context engineering, and reliable AI coding workflows.
Basic familiarity with AI coding assistants will be helpful, but previous experience implementing memory systems is not required.
Join us to learn how shared, searchable memory can help coding agents retain project knowledge, reuse lessons from previous sessions, and avoid repeating the same mistakes