Shlok Khemani
Bio
Shlok Khemani builds and writes about AI memory systems. He's worked on memory and agentic harnesses for startups like Interaction, Zamana, and Mem0, and writes widely about memory system design. His work focuses on the practical and product side of consumer AI memory.
Session (1)
Shlok Khemani is a hands-on builder deeply focused on AI memory systems and human-AI interfaces, having shipped multiple open-source projects (OpenPoke, claude-memory-tools, Centaur) and written extensively on how the major AI assistants approach persistent memory. His session is worth attending for practical, code-level insight into building memory-enabled AI products from someone who has shipped them and written about the design tradeoffs in detail.
GitHub
@shlokkhemani- shlokkhemani/openpoke★ 498Open source implementation of Poke — multi-agent assistant with email triage, reminders, and persistent agents
- shlokkhemani/claude-memory-tools★ 53Example implementations of Claude's Memory Tool API — Next.js web app and Python CLI for building applications with persistent memory
- shlokkhemani/ode-to-yosemite★ 35A to-scale, explorable 3D Yosemite Valley in the browser built end-to-end by Claude Code
- shlokkhemani/centaur★ 9Browser-based UI companion for terminal coding agents using MCP
- zamana-inc/vajraorg★ 50Background coding agent — polls Linear, runs DOT graph workflows, ships PRs
Recent writing (5)
- Google Has Your Data. Gemini Barely Uses It. · blog · Nov 2025
- OpenPoke: Recreating Poke's Architecture · blog · Sep 2025
- Claude Memory: A Different Philosophy · blog · Sep 2025
- ChatGPT Memory and the Bitter Lesson · blog · Sep 2025
- 7 Lessons from Launching My First AI Product · blog · Jul 2025
Podcasts & interviews (1)
- AI in the AM — Week 2 Highlights (June 2026) · The Cognitive Revolution · Jun 2026
Public activity researched automatically · as of Jun 2026