Page 59
12 articlesPersona Overlays: When One Agent Needs Many Voices for Different Customer Cohorts
Shipping one agent persona to a cohort-spanning customer base costs renewals quietly. The fix is overlays, not forks — and slice-level evals that catch the regressions an aggregate score will hide.
The PRD for an AI Feature: Why Your Old Template Misses the Cliff
Deterministic PRDs have no field for what the model gets wrong, what eval score gates ship, or who owns the system prompt. Four sections that fix it.
The Pre-Launch Blast Radius Inventory: The Document Your Agent Team Forgot to Write
Most agent teams discover the absence of a blast radius inventory during their first incident. Here's the artifact, the columns it needs, and how to make it a CI merge gate so it stays accurate.
The Preprocessing Bottleneck That Kills AI Pipeline Throughput
Most teams profile the LLM call and declare victory. The real throughput killer is everything before the model sees a single token — parsing, chunking, embedding, and enrichment that quietly dominate end-to-end latency.
The Promotion Packet for AI Engineers Who Didn't Ship a Feature
How AI engineers whose work is prevention — eval regressions caught, costs that didn't materialize, incidents that never happened — should write the promotion packet a calibration committee can actually score.
The 'What Changed' Query Is the RAG Question Your Index Can't Answer
Vector similarity retrieval treats two versions of the same document as nearly identical, so any RAG system on evolving documents silently hallucinates 'nothing changed' on delta queries. Here is why the failure is structural and what an index that knows about change actually looks like.
The Avatar in the Conference Call: Engineering Real-Time Talking-Head AI for Video Meetings
A 45ms audio-video offset is the threshold where humans flag a talking-head AI as fake. Inside the real-time engineering — viseme schedules, audio master clocks, and the failure modes that never show up in offline eval.
The Reply-All That Wasn't: Agent Outbound Fan-Out Hazards
Agents have no muscle memory for Reply All. When a send tool's recipient field accepts a distribution list indistinguishably from an individual, the planner picks the loudest door. Four practices to keep blast radius bounded.
Retiring an Agent Tool the Planner Learned to Depend On
Removing a function from your agent's tool catalog isn't a syntactic change — it's a behavioral migration. The staged retirement pattern that prevents fallback hallucinations and silent regressions.
Runtime Prompt Hot-Reload: Why Your Prompts Shouldn't Be Locked Behind a Build
Coupling prompt changes to your deploy pipeline is a self-imposed constraint. The runtime hot-reload pattern, its safety primitives, and the failure modes nobody plans for.
The Shadow AI Governance Problem: Why Banning Personal AI Accounts Makes Security Worse
Blocking employees from personal ChatGPT or Claude accounts doesn't stop AI use — it makes it invisible. Here's how to survey shadow AI, build sanctioned channels, and avoid governance theater.
The SIEM Bill Your AI Feature Forgot to Include
Shipping an AI feature multiplies your audit-log volume 10–50x. The SIEM renewal arrives later, with broken detection rules and a legal-hold problem nobody scoped.
About Tian Pan
I'm Tian Pan, an engineer-founder focused on agentic engineering — building autonomous AI systems and scaling engineering teams. I write practical guides on system design, technical leadership, and shipping with AI agents. Previously an early engineer at Uber, Brex, and IoTeX.
