AI & Machine Learning Briefing
Every morning, before you walk into your AI research desk
Lumis has already read arXiv, Hacker News, and every major AI lab blog. Here's what today looked like.
Generate your own AI/ML briefing →Friday, September 11, 2026
Today's top 3
#1
Cognition's SWE-2 claims parity with Fable 5.1 and GPT-Astra on coding tasks
RELEASE
A new frontier coding model entering the ring against OpenAI/Anthropic-class systems raises the bar for agentic software engineering benchmarks that builders rely on for autonomous dev workflows.
#2
OpenAI ships dedicated Agents API, formalizing multi-step tool-use orchestration
RELEASE
A first-party agents endpoint signals OpenAI standardizing agent state, tool calls, and memory—critical for engineers currently hand-rolling orchestration layers with LangChain or custom stacks.
#3
Mathematicians question whether OpenAI can be trusted with unpublished proofs
POLICY
Researchers debating data confidentiality when using frontier LLMs on novel math exposes a real risk for any lab or team feeding proprietary research into third-party APIs.
Also today
Anthropic's September threat report details real-world AI misuse patterns
POLICY
Concrete case studies of adversarial prompting and abuse give safety teams and red-teamers fresh attack patterns to test against their own deployed agents and models.
Google to buy half a nuclear plant's output, underscoring AI's power crunch
MARKET
Direct nuclear procurement by a hyperscaler signals compute-energy constraints are now a first-order variable in training/inference roadmaps for large model builders.
PlanetScale's Neki brings sharded Postgres to scale agent memory and RAG stores
RELEASE
Agent builders hitting single-node Postgres limits for vector/context storage get a new horizontally-scalable option purpose-built for high-throughput AI workloads.
This is today's briefing. Browse the archive →
2 researchers & builders · Founding Member · 500 Seats
Get this every morning. $9/month forever.
Personalized to your sources. Standard price goes to $15/mo after launch.
Claim Founding Seat — $9/mo Forever →
Guaranteed rate. Cancel anytime. No tricks.
arXiv cs.LG + cs.AI daily
Latest research papers ranked by relevance to practitioners, not citation count.
Lab releases & announcements
OpenAI, Anthropic, Google DeepMind, and Meta AI — every significant release, same morning.
HN signal filter
Only the AI/ML threads worth reading, with community sentiment surfaced.
Agent & infra layer
Framework updates, inference benchmarks, and deployment patterns for builders.
Other verticals