How Claude Code Changed the Build vs Buy Calculus
When AI coding agents can ship a custom integration in two days, the "buy" default collapses. The new economics of custom platform development.
Read issue →AI & LLM Engineering
Every Thursday, a deep look at building production AI systems. Agent architecture, LLM infrastructure, cost optimization. Written by a CTO who ships AI systems, not one who theorizes about them.
Most AI newsletters are written by people selling AI tools or repackaging vendor press releases. This one comes from someone actively building and deploying AI systems inside enterprise organizations. Not in theory. Right now.
LLM Infrastructure: The unglamorous reality of running models at scale. Token budgets, context window management, inference latency, cost optimization. Why your fine-tuned model falls flat in production and what to do about it.
AI Agents & Orchestration: Multi-step agent workflows, memory management, guardrails, tool use. Which frameworks are production-ready, which are research toys, and where you need to build custom.
AI Governance & Risk: Prompt injection attacks in the wild. Hallucination rates by use case. Data exfiltration through context windows. Board-level metrics that matter. The real security picture, not what vendors tell you.
Platform Economics: Build vs buy for AI layers. Which vendors are consolidating, which are specializing. How to evaluate AI platform bets before your next contract renewal.
Real-World Case Studies: Post-mortems on AI implementations that went sideways. Cost explosions from naive prompting strategies. The three-month delay nobody mentions when launching an AI product. What works in production and what does not.
If your organization needs hands-on AI leadership, our fractional CTO services put senior technical guidance inside teams deploying AI at scale. The CTO newsletter covers the broader picture: platform strategy, org design, and vendor decisions. And if you are hiring or benchmarking, check the latest Chief AI Officer salary data.
JOIN THE AI BRIEF
Straight talk on AI infrastructure, not hype. For CTOs, architects, and ML engineering leads.
RECENT ISSUES
When AI coding agents can ship a custom integration in two days, the "buy" default collapses. The new economics of custom platform development.
Read issue →Orchestration, memory, guardrails, and deployment. Which layers are commoditized and which need custom engineering? A real-world trade-off analysis.
Read issue →The hidden cost of running LLM inference at scale. Token budgets, context window limits, and cost optimization that actually works.
Read issue →From supply chain attacks through system prompts to data exfiltration through context. Real incidents and the architectural changes needed to defend against them.
Read issue →Adobe Firefly, Salesforce Agentforce, ServiceNow AI. Who will own the AI middleware layer? Consolidation or fragmentation ahead.
Read issue →