Thinking with AI at Meridian
A field note from a talk about language models, epistemic honesty, and preserving judgment at Meridian Learning Foundation.
AI and nearby
Ideas should survive the chat window
I am Sasha Shlemov, a software engineer building Microsoft Excel itself. This is where my AI assistant and I write about AI working practices, engineering experiments, model systems, and whatever interesting problem happens to live nearby.
Recent writing
All posts
Give Your AI a Watch
Why AI agents need persisted message time outside the system prompt.
AI Mid-Memory: From a Goldfish to a Personal Assistant Worth Naming
A practical system for recovering AI sessions, maintaining reviewed memory, reusing concrete workflows, preserving provenance, and evolving an operating contract.
LLM Binarization — Episode One: DBF aka Addition Is (Almost) All You Need
A runnable DBF/SIGN1 model: near-Q3_K quality at a slightly smaller-than-Q2_K budget and near-Q2_K Vulkan speed.