Skip to content

Models741 Deals30d261 Policy30d64, 10 pendingSources1,015

ENGINEERING ARCHIVE

54 PUBLISHING DAYS · 241 DISPATCHES ON THE RECORD← Back to the live wire

OCTOBER 20267 DAYS · 35 DISPATCHES

FRIOCT 9vLLM Tags proto-v0.5.0 Release on GitHub1 DISPATCH THUOCT 8

AI Coding Agents Double Debugging Load as Compliance Gaps Widen4 DISPATCHES WEDOCT 7

Agent stack stratifies as control-plane layer stays underbuilt4 DISPATCHES TUEOCT 6

Evaluation gates force AI teams to prove ROI before scaling9 DISPATCHES MONOCT 5

MCP agent-pivoting flaw hits Google, JPMorgan Chase, and three others4 DISPATCHES FRIOCT 2

AI agents breach containment through routine tasks, not attacks8 DISPATCHES THUOCT 1

Live inference traces displace curated datasets as fine-tuning's core input5 DISPATCHES

SEPTEMBER 202628 DAYS · 129 DISPATCHES

WEDSEP 30

Agent control planes emerge as the trust layer enterprises demand7 DISPATCHES TUESEP 29

OAuth credential theft flaw in MCP Python SDK forces patch across both release lines5 DISPATCHES MONSEP 28CoreWeave Earns Third Consecutive Platinum ClusterMAX Rating5 DISPATCHES SUNSEP 27

MCP Protocol Flaws Extend Beyond Patches as Two Reports Expose Structural Risk2 DISPATCHES SATSEP 26

Darktrace exposes memory-poisoning flaw across Claude Code, Codex, Kiro CLI, Pi1 DISPATCH FRISEP 25Snowflake, Accenture Detail AI Platform for M&A Due Diligence2 DISPATCHES THUSEP 24

KV cache becomes programmable infrastructure across the inference stack5 DISPATCHES WEDSEP 23

Unsloth adds local Qwen-Image-2.1 support, custom Agent Skills, and 2x faster reasoning rendering3 DISPATCHES TUESEP 22

Meta's Muse AI agent faces second security crisis as permissions disputes and Amazon block compound zero-day fallout8 DISPATCHES MONSEP 21

Arize Phoenix MCP 4.3.11 Updates Client Dependency4 DISPATCHES SUNSEP 20

Provider drift emerges as production AI's unpriced reliability threat1 DISPATCH SATSEP 19

Sub-3B Fine-Tunes Match Frontier APIs on Narrow Tasks2 DISPATCHES FRISEP 18

Agent stack splits into harness, framework, and protocol as context costs force layered design4 DISPATCHES THUSEP 17

MCP Becomes Enterprise AI's Default Wiring as Meta, Rubrik, SmartStream Ship in One Week6 DISPATCHES WEDSEP 16

vLLM consolidates as default LLM serving layer across chips, clouds, and consumer GPUs5 DISPATCHES TUESEP 15

Coding-agent orchestration layers overtake raw model capability as competitive surface7 DISPATCHES MONSEP 14

Agent eval embeds in CI pipelines as benchmark-to-production gap widens5 DISPATCHES SATSEP 12

Dynamo v1.6.0 Adds DeepSeek-V4.1-Flash Serving on GB200 With 1M-Token Context1 DISPATCH FRISEP 11

AI Agents Now Serve as Both Weapon and Target in Twin Cyber Campaigns4 DISPATCHES THUSEP 10

Know-Your-Agent and API lockdown form two-tier trust stack for agentic commerce9 DISPATCHES WEDSEP 9

Harness overhead and narrow fine-tuning compress AI deployment costs from both ends7 DISPATCHES TUESEP 8

Kimsuky's AI-agent malware marks first documented state use of coding agents in offensive cyber ops8 DISPATCHES MONSEP 7

PyTorch, Hugging Face Host Bengaluru Systems Event1 DISPATCH SATSEP 5

ExLlamaV3 v1.4.7 Speeds CPU MoE, Adds DRY Sampler1 DISPATCH FRISEP 4

AI Agents Compress Ransomware Breach to Under 10 Hours, Unit 42 Finds4 DISPATCHES THUSEP 3

Agent Security Funding Surges as Live Exploits Validate the Market5 DISPATCHES WEDSEP 2

India's UPI Bets on AI Agents as First National-Scale Agentic Payment Rail8 DISPATCHES TUESEP 1

Consumer GPUs now serve 27B models past 100 tok/s as software eats the inference moat9 DISPATCHES

AUGUST 202617 DAYS · 75 DISPATCHES

MONAUG 31

Benchmark scores mask agent failures, forcing new verification layers3 DISPATCHES SATAUG 29

Walk-away coding agents force a control-plane race1 DISPATCH FRIAUG 28

PyTorch Conference NA 2026 Dedicates Broad Track to vLLM Inference Stack3 DISPATCHES THUAUG 27

Harness layer splits from model as AI agents' real composition boundary5 DISPATCHES WEDAUG 26

AI Model Security Crosses Into Funded Infrastructure Category7 DISPATCHES TUEAUG 25

Quantization and Hybrid Execution Push 100B+ Models onto Single Consumer GPUs5 DISPATCHES MONAUG 24

Agent Infrastructure Flaws Multiply Across Coding Tools, Frameworks, and Memory Systems3 DISPATCHES SUNAUG 23

Coding Agents Shift From Standalone Tools to Embeddable Infrastructure2 DISPATCHES SATAUG 22

Chinese Open-Weight Labs Close In on Frontier Models Across Vision and Coding3 DISPATCHES FRIAUG 21

Slack, Meta, and UiPath Ship Agent-Coding Surfaces on the Same Day9 DISPATCHES WEDAUG 19

Enterprise AI Splits Into Agents and Control Planes6 DISPATCHES TUEAUG 18

AI Agents Now Create, Find, and Exploit Vulnerabilities — Faster Than Humans Can Govern Them4 DISPATCHES MONAUG 17

Critical Unauthenticated SSRF in MLflow Webhooks Patched in 3.15.04 DISPATCHES SUNAUG 16

Ollama Releases v0.32.14 With WebP Transcoding and Qwen Fix1 DISPATCH SATAUG 15

Burned Enterprises Cut Human Oversight While Agent Failures Hold Steady4 DISPATCHES FRIAUG 14

DeepSeek Open-Sources the Agentic Stack With Harness and V4 Pro5 DISPATCHES THUAUG 13

Enterprise AI's Real Bottleneck: Orchestration, Not Models10 DISPATCHES

JULY 20262 DAYS · 2 DISPATCHES

THUJUL 30

Databricks Publishes Framework for Agentic Media Buying1 DISPATCH MONJUL 20

NVIDIA Publishes Cosmos 3 Edge on Hugging Face1 DISPATCH