◆ DAILY BRIEFING
Thursday, March 19, 2026
-
Engineer OpenAI's Codex architecture disclosure reveals MCP failed for production agentic workflows — they abandoned it and built a custom bidirectional JSON-RPC protocol because MCP can't handle streaming, approval flows, or structured diffs.
OpenAI's Codex team abandoned MCP for production agent workflows and discovered that non-deterministic tool ordering silently destroys prompt cache hits — if you're building agentic systems, audit you…
Read full briefing → -
Security Three nation-state toolkits dropped simultaneously with published IOCs: Lazarus planted a typosquat of Meta's react-refresh (42M weekly downloads) on npm delivering PylangGhost RAT, APT28's entire C2 infrastructure leaked revealing 2,800+ exfiltrated emails and 140+ persistent Sieve forwarding rules across six countries, and a second iOS exploit kit — DarkSword — puts 270M unpatched iPhones at risk using repurposed U.S.
Three nation-state toolkits were exposed in a single cycle — Lazarus poisoning npm, APT28 exfiltrating thousands of emails via webmail XSS, and DarkSword targeting 270 million unpatched iPhones with r…
Read full briefing → -
Data Science GPT-5.4 nano just landed at $0.20/M input tokens — 5 million classifications for $1 — while OpenAI's own Codex architecture teardown simultaneously reveals that a non-deterministic tool-ordering bug silently destroyed their prompt cache, 10x-ing per-request compute with zero functional test failures.
GPT-5.4 nano at $0.20/M tokens reprices the inference floor — 5 million classifications for $1 — but OpenAI's own Codex teardown reveals that a non-deterministic tool-ordering bug silently 10x'd their…
Read full briefing → -
Product OpenAI declared internal 'code red' over Anthropic's enterprise dominance and is killing Sora, its browser, hardware, and ad experiments to refocus entirely on coding tools and business workflows — while Microsoft's Copilot has penetrated just 3% of Office subscribers and chose Anthropic's Claude (not GPT) to power its new Cowork agent.
The enterprise AI market hit a structural inflection point this week: OpenAI declared 'code red' and killed consumer experiments to chase Anthropic's enterprise lead, Microsoft's Copilot stalled at 3%…
Read full briefing → -
Leader JPMorgan pulled a $5.3B Qualtrics debt deal because investors refuse to buy SaaS paper in an AI-disruption environment — the first time AI anxiety has killed a major financing at the credit-market level.
JPMorgan killing a $5.3B SaaS debt deal on AI disruption anxiety is the moment AI risk crossed from boardroom speculation into credit-market pricing — and it's happening the same week OpenAI declared…
Read full briefing → -
Investor UTIMCO's latest fund disclosures reveal the most extreme return concentration in VC history: three LLM companies' gross profit now equals ~70% of all VC profits from the prior decade — and 100% of it is unrealized paper gains.
VC's greatest returns in history — 70% of a decade's profits concentrated in three LLM companies — are 100% unrealized paper gains, the credit market just started pricing AI disruption into SaaS debt…
Read full briefing →