AI Plugins for Microsoft 365 and Google Workspace
Patterns for shipping AI features inside Microsoft 365 and Google Workspace in 2026 — Add-Ins, Copilot Extensions, and Workspace integrations.
Agentic AI, LLM engineering, and the models behind modern automation — multi-agent systems, LLM evaluation and comparisons, RAG, fine-tuning, AI infrastructure, security, and production AI engineering.
From the blog
Patterns for shipping AI features inside Microsoft 365 and Google Workspace in 2026 — Add-Ins, Copilot Extensions, and Workspace integrations.
The architectural transitions that take an AI project from a PoC to production-grade in 2026 — and the things teams routinely miss.
The evolution of attention from the original transformer to 2026's multi-query and grouped-query variants — what changed and why it matters.
How agents convert vague human goals into executable steps in 2026. The decomposition patterns and the failure modes that derail them.
Production agents that surface uncertainty cleanly are dramatically more useful than confident-but-wrong ones. The 2026 uncertainty-design patterns.
Multi-layer cache designs for AI apps — prompt cache, response cache, retrieval cache, embedding cache — and how they compose in 2026.
California's AB 2013 forced training-data disclosure for frontier model providers. What is now public, what is not, and what other states are following.
Canarying new model versions catches regressions early. The 2026 patterns for safe LLM canary deploys and rollback automation.
ToT, GoT, and self-consistency are CoT successors. The 2026 head-to-head comparison and where each pays its compute cost back.