Blog
Notes on building, evaluating, and deploying Ember.
Shadow AI Is Already Inside the Enterprise. The Answer May Not Be to Ban It.
9/27/2026
How a governed AI Workbench can give employees the freedom to use AI while giving organizations control over where their knowledge goes
The Next Connector Catalogue Will Be Built by Builders
9/27/2026
No vendor will build every AI integration needed by every regional business. A governed Builders Programme can help software houses, students and solo developers turn existing APIs and webhooks into tested, portable MCP capabilities that Ember and other assistants can use safely.
A Method Is Not an Agent Graph
9/27/2026
Are enterprise AI Methods just agent graphs? Not quite. A Method defines the governed procedure, evidence, assessment and human decisions. A graph is one possible execution design, and an AI agent is only one possible participant. The distinction becomes especially important when much of the work is performed in external AI and engineering platforms.
The Document-First Principle for Enterprise AI
9/27/2026
Why an AI assistant should prepare evidence, plans, and reviews before it changes operational systems.
One Default AI Model Is Not Enough
9/27/2026
Conversational reasoning, structured output, embeddings, and evaluation are different jobs.
A Private AI Journal Should Belong to the Individual
9/27/2026
Using conversation history to support reflection without turning personal memory into employer surveillance.
Private AI Deployment Is an Operating Model, Not a Docker File
9/27/2026
What organisations take responsibility for when they move AI, data, authentication, and storage into their own cloud.
Agent Harnesses: Operating and Governing Enterprise AI Agents
9/27/2026
A model is not an agent, and an agent is not a harness. Placeholder for an article on the identity, access, context, and provenance controls that turn an AI capability into something an enterprise can actually operate and govern.
Ember Deployment and Server Options
9/27/2026
From Vercel and managed Supabase to Azure, AWS, or a fully private deployment -- how Ember can move beyond its current hosting without slowing product development.
Practical AI Architecture Governance -- and Where Ember Fits
9/27/2026
A new piece on proportionate, evidence-based AI governance -- and a look at how Ember's own Milestone 7 governance pilot could put those same ideas into practice.
Why Enterprise AI May Start with Dedicated Customer Instances
9/27/2026
Shared multi-tenant SaaS is not always the right first architecture for AI products handling sensitive governance evidence. Placeholder for a vendor-neutral discussion of dedicated-instance-first deployment and the SaaS-ready path underneath it.
Practical Architecture Governance in Southeast Asia
9/27/2026
Enterprise architecture governance does not have to mean importing a large framework and applying every part of it. For organisations across Southeast Asia, the better question is: what is the smallest governance system that will produce clear decisions, credible evidence, and responsible change?
Your AI Assistant Should Tell You Which Model Answered
9/27/2026
Why model identity, message-level provenance, and transparent switching matter in enterprise AI.
Before the AI-Native SDLC: Test What Is Worth Building
9/27/2026
Faster AI-assisted coding does not make an untested assumption more reliable -- it can just help an organisation build the wrong thing sooner. An independent response to Anthropic's AI-Native SDLC Playbook on the experimental layer that should sit upstream of it.
Conversation History Is Not the Same as AI Memory
9/27/2026
Designing short-term context, conversation continuity, and long-term memory without confusing users.
Why AI Responses Need Structure, Not Just Markdown
9/27/2026
Separating conversation, summaries, next actions, citations, documents, and trusted links.