Why Agents Fail in Production
The failure modes that never show up in a demo, and what they have in common.
Evaluating Agents Without a Benchmark
How to know an agent is getting better when there is no leaderboard for your org.
The Cost of a Wrong Write
Why confirmation gates belong on some actions and nowhere near others.
Beyond the Vector DB
Why retrieval quality depends more on your schema than your embedding model.
The path to AGI
Notes on what actually compounds in agent capability, and what does not.
ReAct vs Plan-and-Execute
Choosing the right agent loop for Agentforce workloads.
TypeScript vs Python
Picking a runtime for agent tooling when both teams have opinions.
Context Windows Are Not Memory
A large context window solves a different problem than the one teams reach for it to solve.
Tool Design Is the Real Prompt
Most agent behaviour is determined before the model reads a single instruction.
The End of Prompt Engineering
Prompts are becoming interfaces. What replaces the prompt as the unit of work.
When Not to Use an Agent
A short list of problems where a flow, a formula, or a report is the better answer.
Human in the Loop Is a Design Decision
Where to put the person, and why the default answer is usually wrong.
Observability for Agent Systems
What to log so that the question after an incident has an answer.
Multi Agent Systems Are Usually One Agent
When splitting into several agents helps, and when it is one agent with extra latency.
Data Hygiene Is an AI Problem Now
The CRM cleanup nobody funded is suddenly the thing blocking the agent rollout.