Notes
from the team.

A practical guide to AI enablement, AI adoption, and reusable agent skills. See how Brains, Sessions, Skills, and Rules support an engineering team’s AI coding workflow.

How Telexpress used XTrace to diagnose where its AI agents were going wrong, turn its own bug history into 24 enforced rules, and close the month with zero violations in audit.

Your AGENTS.md is too long when agents skip rules. Why ~150 lines is the real budget in 2026, what to keep, and how to move the rest into checks that run.

Coding agent observability in 2026: traces, sessions, tool calls, CLAUDE.md drift and the token-to-ticket ratio for Claude Code, Cursor and Codex spend.

What agent procedural memory is, in plain English — why agents need memory outside the model, what a stored procedure contains, and how it differs from semantic memory.

Too many CLAUDE.md, MEMORY.md and skill files? When your agent's skills folder outgrows its context window, there are four fixes: consolidate, restructure, split rules from procedures, or stop reading files and start retrieving them.

Foundation models are commoditizing fast, so why don't they wipe out vertical AI startups? Because the moat was never the model. Here are the three levers that make vertical and departmental AI defensible, and how to compound them into something the foundation labs can't copy.

The feature we are launching this week is building your own LLM-Wiki without coding or using Claude Code. We extract AI memory and context from your chat history, store it in an encrypted vector database, and organize it in a file system that Obsidian prefers. Now you can keep building your second brain without any cold start!

Memory isn't a search problem. It's a state management problem. Here's how X-Mem tracks what your agent knows, what it used to know, and why those two can disagree.

Mem0, Zep, GBrain, Karpathy's wiki, Hindsight: what each one actually does, and the gap they all leave open.

Most Ai agents don't have memory, they have a list, which isn't a memory system. Why context degradation breaks long-running AI agents and what belief revision fixes.

Learn how to export Gemini conversations step-by-step, back up your chat history, and keep AI conversations reusable across tools.

XTrace is the first production-ready encrypted vector database SDK that performs nearest-neighbor search on fully encrypted data, so the server never sees your vectors, queries, or documents. Built for healthcare, legal, enterprise R&D, and AI agents handling sensitive context.

Anthropic just announced a model it calls "by far the most powerful AI we've ever developed." No public release. No waitlist. No pricing page.

Learn how to set up XTrace private vector database

Claude's memory feature is now free for all users, but it still has structural limits that no subscription tier can fix. Here's what it actually does, where it falls short, and how to get persistent memory that works across every AI tool you use.

Learn how Claude's memory actually works, how to view and manage what it stores, and how to build persistence that lasts beyond a single session.

Learn how to view, edit, and delete what Gemini remembers about you and how to decide how much you actually want it to know.

Learn how to view, edit, and delete what ChatGPT remembers about you, and how to decide how much you actually want it to know.

Your AI agent doesn't need total recall, it needs to know what it can't afford to forget. The most dangerous mistakes come from context it never thinks to ask for.

Learn how to export Claude conversations step-by-step, back up your chat history, and keep AI conversations reusable across tools.

Your AI already knows how you think. But that understanding is scattered across every tool you use. What happens when it all connects?

Learn how to export ChatGPT conversations step-by-step, back up your chat history, and keep AI conversations reusable across tools.

AI agent handoff often loses decisions, reasoning, and evidence. Learn how structured memory enables reliable multi-agent workflows.

We've solved storage and retrieval. But context still dies when work moves between tools, agents, or teammates. The AI Context Bus is the missing layer that lets memory flow across workflows.

Forward deployed engineers scale expertise manually. AI agents can scale it automatically, but only if context moves with them. Here’s what’s missing.

Your AI vendor isn’t just hosting your data, they’re holding your company’s competitive edge hostage. Here’s how to break the cycle.

Authentication won't cut it in the AI era. Your agents need a context login

Stop acting as the manual bridge between your AI agents and learn how automated context switching lets you scale

RAG is great for reading a library, but it can’t write an autobiography. Your AI agents are forgetting their instructions, here's how to transform from static retrieval to dynamic memory management.