context window management
- AI Strategy (315)
- Generative AI (106)
- IT Leadership (33)
- Leadership (76)
- Product Management (344)
- Product Management Leadership (306)
- Uncategorized (32)
-
Claude Fable 5.1: A Practical Adoption Guide for AI Leaders
A practical framework for deciding where Claude Fable 5.1 belongs, how to test its judgment, model whole-run cost, and govern sensitive use.
-
How to Build a Production-Ready AI Coding Workflow
A practical operating model for turning AI-generated patches into scoped, reviewable, tested, and reversible changes your engineering team can ship.
-
Persistent Memory for AI Assistants: A Product Playbook
A practical guide to deciding what an AI assistant should remember, operating its memory lifecycle, and measuring whether users can trust it.
-
How to Give Autonomous Agents Context and Permission to Act
A practical operating model for agents that detect work, use current context, take bounded action, and escalate before consequences outrun control.
-
Persistent Memory for AI Agents: A Product Leader’s Guide
A practical framework for deciding what an AI agent should remember, how it should retrieve and forget memories, and how to test the value safely.
-
Chinese Frontier AI Models: A Product Leader’s Decision Guide
A practical framework for choosing Kimi K3, Qwen 3.8 Max, or DeepSeek V4 Flash by completion reliability, accepted-result cost, and product fit.
-
Context Engineering: How to Build Reliable AI Applications
A practical framework for designing context, diagnosing agent failures, budgeting each turn, and evaluating AI applications before production.
-
How to Design Persistent Memory for Reliable AI Agents
A practical framework for deciding what an AI agent should remember, retrieving it safely, correcting stale facts, and proving that memory helps.
-
AI Agent Skill Overload: How to Curate What Stays
Learn how to test, approve, fork, and retire AI agent skills before instruction conflicts and hidden assumptions weaken your team’s output.
-
Persistent AI Agent Memory: A Knowledge Graph Blueprint
A practical blueprint for modeling, governing, retrieving, and evaluating knowledge graph memory that an AI agent can safely use over time.
-
How to Build Persistent Business Memory with AI Knowledge Graphs
Learn how to design a time-aware, evidence-linked knowledge graph that lets AI recover current decisions, their history, and provenance.
-
Codex Token Cost Optimization Without Losing Useful Context
A practical model for trimming Codex context, preserving decisions, and proving token savings without creating more retries or review work.
-
Claude Code for Non-Technical Professionals: A Safe Start
A practical, low-risk way to use Claude Code for file-based product work, with a bounded first project, review gates, and reusable prompts.
-
Reliable AI Coding Requires Four Kinds of Control
A practical framework for controlling requirements, context, verification, permissions, and recovery when teams build software with AI agents.
-
From Brain Dump to Done: How Todoist’s Ramble Captures Tasks in Real Time with AI
Ramble turns stream-of-consciousness voice notes into structured tasks in real time by skipping transcription and using a Gemini live audio model. The Doist team pairs instant…
-
Never Lose Your AI Superpowers: How I Sync Context and Skills Across Every Device
Sharing AI context files and skills across devices—and with teammates—shouldn’t be painful. In this first-person walkthrough, I break down the real-world pitfalls I hit with Dropbox,…
-
Inside Banani: How a Canvas-First AI Designer Elevates UX and Accelerates Product Teams
Banani’s canvas-first AI designer reframes how teams access high-quality UX by generating design—not just code. I break down how the team handles parallel edits, per-screen context…
-
5 powerful ways I use Pendo MCP to bring product analytics into ChatGPT, Claude, and Cursor
Pendo MCP brings product analytics directly into Claude, ChatGPT, and Cursor, so insights follow me wherever I work. I can triage anomalies, prep discovery with behavioral…
-
Agentic Architecture Demystified: How Modern AI Systems Plan, Learn, and Execute at Scale
Modern AI isn’t just a model; it’s an agentic system that plans, retrieves, acts, and evaluates in a tight loop. In this piece, I break down…
-
Multi‑Agent Systems Demystified: Why One AI Isn’t Enough—and How I Ship Faster With Many
Multi-agent systems turn a single clever model into a reliable product engine by coordinating specialized agents—planner, executor, and verifier—inside robust AI workflows. I use a retrieval-first…
Weekly digest
One email a week on AI products, enablement and hiring. No fluff.
Browse topics
- AI Strategy (315)
- Generative AI (106)
- IT Leadership (33)
- Leadership (76)
- Product Management (344)
- Product Management Leadership (306)
- Uncategorized (32)
Work with me
45-minute consultation on AI product strategy, GTM and PM hiring — no charge.