Skip to main content
Back

Grok Bot joins Slack teams; Gemini 4 Carbon runs days

Google releases Gemini 4 Carbon, Grok Bot ships Slack AI teammates, FAISS launches 10x faster clustering, Anthropic reads AI inner monologue

November 12, 2026 6 min read

Hey,

AI is quietly moving from your assistant to our assistant. Grok Bot joins team Slack channels. Gemini 4 Carbon runs solo for days on hard problems. And NYU just showed you can teach a model new skills by deleting 0.05% of its weights.

The pattern? Less is becoming more. Shared context, fewer weights, longer autonomy.

We're entering the era of AI that works with teams, not just for individuals. That's a bigger shift than it sounds.
In Partnership with · partner_image
Author photo
Today's Author
Lior Alexander. Founder of AlphaSignal and former ML Engineer. Previously built ML systems at Iguazio, Guesty, Enphase, Mila.
Summary

Read time: 6 min 39 sec

Top News

▸ Grok Bot 1.0 ships AI teammates that join Slack, Teams, and meetings

Unblocked

▸ A Pile of MCP Connectors Is Not a Context Engine

Top News

▸ Google releases Gemini 4 Carbon, its most powerful model for complex agentic work

Datadog

▸ Ship LLM features faster, with less guesswork

Top Repo

▸ FAISS ships e-means, a clustering algorithm 10x faster than k-means

Signals

1. Google DeepMind ships Gemini Robotics 2.5 with 96% accuracy sorting car parts
2. Wiz: One-page guide breaks down how AI agents operate and where security risks show up in each step
3. NYU method matches full fine-tuning by deleting just 0.05% of LLM weights
4. Anthropic tool reads AI "inner monologue" , catches self-preservation thoughts before deletion
5. Zyphra releases open-source voice cloning model that runs in your browser
6. Unsloth AI ships quantized Qwen3 models that run on 12GB RAM with 5–10% accuracy gains
Our Events
From Vulnerable Dependencies to Tested Pull Requests with Snyk
Dec 9, 10a PT · Online · Register now
Top News
Grok Bot 1.0 ships AI teammates that join Slack, Teams, and meetings
11,994 Likes
Grok Bot 1.0 ships AI teammates that join Slack, T

Grok Bot 1.0 just dropped, and it flips the script on how AI assistants work. Until now, your AI was yours alone. Now it belongs to the whole team.

Here is what changed:

  • Team Bots: one shared Bot your whole team can use inside Slack and Microsoft Teams channels
  • Meeting support: invite a Bot to Google Meet, Zoom, or Teams calls and it listens, answers in chat, then sends a summary with action items
  • Proactive work: the Bot spots unfinished tasks and offers to take them on without you asking
  • Shared memory: the Bot learns from everyone it works with, building up team knowledge over time

Setup needs a team admin to approve access. What the Bot learns from teammates also needs owner approval before it sticks. So there are guardrails.

The big unlock here is that your team can now give a Bot real ongoing work across support, sales, or accounting, and it keeps context across the whole group, not just one person.

Read more
forward →
Presented by Unblocked
A Pile of MCP Connectors Is Not a Context Engine

Your team connected Slack, Jira and GitHub over MCP, and engineers still spend their time steering agents to the right answer.

Unblocked's benchmark shows why: connectors fetch from one tool at a time, so every agent stitches the pieces together itself, trusts the first doc that looks right, and adds a search to your token bill each time.

Read what changes when a context engine sits on top of your connectors: the same coding task finished 83% faster at 60% lower cost.

Read the article
partner with us →
Top News
Google releases Gemini 4 Carbon, its most powerful model for complex agentic work
8,425 Likes
Google releases Gemini 4 Carbon, its most powerful

Google just dropped Gemini 4 Carbon, their strongest model yet. Think of it as Argon's smarter sibling that takes more time to think before answering.

Here's the key idea: Carbon uses extra internal reasoning steps before giving you a response. More thinking equals better answers on hard problems. Google's own teams have seen Carbon run for days on complex engineering tasks without giving up.

What it's built for:

  • Multi-step debugging across large codebases
  • Cybersecurity work like finding and patching vulnerabilities
  • Long-running agentic tasks that take hours or days
  • Open-ended scientific research with a 1M token context window

Pricing starts at $2/$10 per 1M input/output tokens, with a 95% discount on cached inputs. Already using Argon? Switching is literally a one-word model ID change: gemini-4-carbon. Access it today via Google AI Studio or the Gemini API.

Read more
forward →
Presented by Datadog
Ship LLM features faster, with less guesswork

Your LLM app works in staging and breaks in production. This guide shows how engineering teams:

  • Trace every prompt, span, and tool call
  • Catch quality regressions before customers do
  • Keep token spend predictable

Ship AI features on your own schedule.

Download the guide
partner with us →
Top Repo
FAISS ships e-means, a clustering algorithm 10x faster than k-means
6,832 Likes
FAISS ships e-means, a clustering algorithm 10x fa

K-means is everywhere. It groups data points into clusters, used in search engines, recommendation systems, and AI model training. But it's slow. Meta FAIR just shipped e-means, a faster drop-in replacement, now live in FAISS 1.16.0.

The core idea: instead of scanning the full dataset every round, e-means reads small chunks and tracks a running average per cluster. Smarter updates, same result, way less work.

Here's what that looks like in practice:

  • 10-45x faster on GPU, 8-15x on CPU vs standard k-means
  • Matches 25-epoch k-means quality in just 1.4 to 2.6 epochs
  • Tested on 26 datasets, from 100K to 10 billion points
  • Zero extra settings to tune beyond k (number of clusters)

Swapping it in takes one line. Replace Kmeans(d, k).train(x) with Emeans(d, k).train(x). That's it. You get better quality clusters, faster, and with no config headaches.

Read more
forward →
Signals
1
Google DeepMind ships Gemini Robotics 2.5 with 96% accuracy sorting car parts
5,532 Likes
2
One-page guide breaks down how AI agents operate and where security risks show up in each step
Presented by Wiz
3
NYU method matches full fine-tuning by deleting just 0.05% of LLM weights
1,760 Likes
4
Anthropic tool reads AI "inner monologue" , catches self-preservation thoughts before deletion
753 Likes
5
Zyphra releases open-source voice cloning model that runs in your browser
4,170 Downloads
6
Unsloth AI ships quantized Qwen3 models that run on 12GB RAM with 5–10% accuracy gains
2,522 Likes
At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 300,000 developers stay informed and ahead. We're passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.

Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We're based in the United States. By using our site, you agree to be governed by U.S. laws.
Looking to promote your company, product, service, or event to 300,000+ AI developers?
Work with us