This Week in All Things AI - Week 20-2026
This Week in All Things AI – Week 20/2026 distills the firehose of updates shared in the “All Things AI” Telegram group from Sunday 10th May to Saturday 16th May 2026 into a structured brief you can scan in a few minutes.
The through‑lines this week are: rapid expansion of frontier model access (OpenGradient, xAI), Anthropic’s continued push into agents, programmatic usage, and verticals like legal and SMB, and a noticeable uptick in agentic content/marketing stacks exemplified by Higgsfield’s Supercomputer.
We open with infra and platform news: OpenGradient’s 13 new frontier models (including Claude Opus 4.7, GPT‑5.4/5.5, o3, Gemini 3.1, Grok 4.20), Anthropic’s Agent View for Claude Code, a new pool of dedicated programmatic credits across paid plans, and Claude for Legal and Small Business with 20+ MCP connectors and practice‑specific plugins. From there, we cover xAI’s Grok Skills and the Grok Build CLI beta for SuperGrok Heavy subscribers, Notion’s developer platform turning workspaces into agent hubs, and Higgsfield’s self‑learning “Supercomputer” that orchestrates multi‑model creative pipelines across social channels. We close with risk/governance and AEO: AI‑transcribed meetings and privilege, Google’s AI Threat Tracker, Elon’s hints about Grok V9, and Google Search’s new guide for optimizing content for generative AI and AI agents.
If you know founders, operators, or builders trying to stay ahead of the curve on models, agents, and real‑world deployments, share this weekly brief so they can scan the most important launches, risks, and resources in a few focused minutes.
The following messages were posted on the ' All Things AI ' Telegram group from Sunday 10th May 2026 to Saturday 16th May 2026
Sunday 10th May 2026
via Robby Yung
Inference on OpenGradient just expanded significantly.
13 new frontier models are now available across the network, including Claude Opus 4.7, GPT-5.4 & 5.5, o3, Gemini 3.1, and the full Grok 4.20 family.
Every inference runs inside TEE-attested environments with cryptographic
Monday 11th May 2026
Corporate lawyers say AI-transcribed meeting notes may not have attorney-client privilege, making otherwise protected conversations discoverable in lawsuits
Google's TIG reports the first confirmed instance of “prominent cybercrime threat actors” using AI to find and weaponize a zero-day in a web-based admin tool
Google Threat Intelligence Group is dropping our latest AI Threat Tracker report today, which covers several threats we are watching through a variety of means. The report includes some details of the first 0day exploit we've found developed with AI. 1/x
cloud.google.com Adversaries Leverage AI for Vulnerability Exploitation, Augmented Operations, and Initial Access \...
Explore GTIG's 2026 report on how adversaries leverage AI for zero-day exploits, autonomous malware, and industrial-scale cyber operations.](https://t.co/klvOrX31xv)
[**Adversaries Leverage AI for Vulnerability Exploitation, Augmented Operations, and Initial Access
Explore GTIG's 2026 report on how adversaries leverage AI for zero-day exploits, autonomous malware, and industrial-scale cyber operations.
https://cloud.google.com\
\
Colossus Magazine Publishes Profile Interview of Cognition AI Co-Founder Scott Wu by Jeremy Stern
Jeremy Stern, editor-in-chief of Colossus, interviewed Scott Wu, co-founder of Cognition AI, covering his competitive programming achievements, founding of the company after his mother's death from lung cancer, and Cognition's AI software engineer Devin achieving a $445 million revenue run rate with customers including the US Army, Goldman Sachs, and Mercedes-Benz. The profile highlights Wu's background as the greatest US competitive programmer with perfect scores in advanced math competitions as a child, and the company's rapid growth toward a $25 billion valuation.
Scott Wu is the co-founder of Cognition AI, one of the fastest-growing companies in history. He’s also the greatest competitive programmer the US has ever produced. You may have seen him doing impossible card tricks and mental math.
You’ve never seen him asked about weed,
[The Wu Tapes: Q&A with Cognition's Scott Wu
Cognition founder Scott Wu on building the AI coding agent Devin, his math-Olympiad gold medals, and what remains for humans in an age of AGI.
https://colossus.com\
\
Tuesday 12th May 2026
Anthropic Launches Agent View for Managing AI Coding Sessions
Anthropic released Agent View in Claude Code version 2.1.139, a dashboard that shows status for all coding sessions—like pinned tasks, ones needing input, or completed checks. Run claude agents to start, dispatch new sessions with a prompt, and switch easily without extra tabs; sessions persist even if you close the terminal. Engineers like Awni Hannun call it 'exceedingly useful—10/10' for daily work, though parallel agents can quickly hit rate limits on Pro or Max plans.
New in Claude Code: agent view.
One list of all your sessions, available today as a research preview.
The constraint Anthropic just removed was the terminal tab.
One Claude session per tab capped your parallelism at whatever context-switching cost you could absorb. For most devs that was one. Maybe two. The model could run forever in parallel. The human couldn't.
Agent view
New in Claude Code: agent view.
One list of all your sessions, available today as a research preview.
This might be more likely useful to others in group members extended family who want to use Anthropic Claude in a better manner. Allmost all members in this group are unlikely to be the target audience for this content but you are welcome to skim it and if you find something useful that you weren't already aware of then consider it as a bonus.
A collection of free guides to master Claude from curator Ruben Hassid whose newsletter I subscribe to.
[**Claude 101**
A collection of free guides to master Claude, from beginner to expert.
https://claude101.com](https://claude101.com/)
I'm not a SuperGrok Heavy subscriber , just a regular SuperGrok one and I'm now able to access Grok Skills
Grok Web just got a massive upgrade with Skills.
This is a huge productivity update for the AI chatbot.
You can now tell Grok things like:
- “Make me a pitch deck”
- “Edit this PDF”
- “Analyze this spreadsheet”
…and it executes the workflow directly inside the chat.
Built-in
Wednesday 13th May 2026
Anthropic just made another serious push into legal.
According to Anthropic, "Legal is now the #1 power-user job function in Claude Cowork."
The release adds 20+ connectors and 12 practice-specific plugins designed to plug Claude into the systems legal teams already use, from Microsoft apps to document management, e-discovery, and research tools.
[**Claude for the legal industry
We're releasing 20+ new MCP connectors that link Claude to the software that the legal industry runs on and 12 new plugins tailored specific legal work and practice areas.
https://claude.com\
[Anthropic Unveils 'Claude for Legal' With 12 New Plugins, 20+ MCP Connectors & More
Claude for Legal is more than just a product launch. It's an expansion of Anthropic's current legal offerings, built in conjunction with major legal tech providers and operating within the existing Claude interface.
https://www.legaltechnologyhub.com\
[Anthropic Goes All-In on Legal, Releasing More Than 20 Connectors and 12 Practice-Area Plugins for Claude
Anthropic today took its biggest step yet into the legal market, releasing more than 20 new MCP connectors linking Claude to the software that law firms and legal departments run on, along with 12 new...
https://www.lawnext.com\
[**GitHub - anthropics/claude-for-legal: A suite of plugins for legal workflows**
A suite of plugins for legal workflows. Contribute to anthropics/claude-for-legal development by creating an account on GitHub.
https://github.com](https://github.com/anthropics/claude-for-legal)
Thursday 14th May 2026
Anthropic Adds Dedicated Credits for Claude's Programmatic Tools
Paid subscribers on Pro, Max, Team, and Enterprise plans will receive credits just for programmatic AI tools like the Claude Agent SDK and third-party apps, keeping chat and coding limits untouched. Credit amounts match plan prices, from $20 for Pro to $200 for Max 20x, claimed once via email and auto-renewing monthly.
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage.
The credit covers usage of:
- Claude Agent SDK
- claude -p
- Claude Code GitHub Actions
- Third-party apps built on the Agent SDK
To add some clarity: you don't pay extra. It's the same subscription, same price per month.
What's new our sub now covers two separate pools:
· Interactive → sub limits, unchanged
· Programmatic → new $20–$200 included(!!) credit, metered at API rates
Starting June 15, paid Claude plans can claim a dedicated monthly credit for programmatic usage.
The credit covers usage of:
- Claude Agent SDK
- claude -p
- Claude Code GitHub Actions
- Third-party apps built on the Agent SDK
Anthropic launches Claude for Small Business, featuring a host of automated services like bookkeeping functions, business insights, and tools for ad campaigns
[Introducing Claude for Small Business
We're launching Claude for Small Business, a package of connectors and ready-to-run workflows that put Claude inside the tools small businesses use every day.
https://www.anthropic.com\
\
via Clayton Lam
Hey, Clayton here from Notion. I just joined the landing GTM team for SG / HK. Nice to meet you all.
Some exciting news from us today, we just launched our Developer Platform that allows clients to:
sync any data into Notion workspaces (think Salesforce, Zendesk, Shopify)
Build any agent tool
Orchestrate any agent in Notion (invite Claude, Manus, Cursor etc. to work in Notion)...
[Notion just turned its workspace into a hub for AI agents
| TechCrunch
Notion's new developer platform lets teams connect AI agents, external data sources, and custom code directly into their workspace as the company pushes deeper into agentic productivity software.
https://techcrunch.com\
\
Friday 15th May 2026
via xAI
An early beta of Grok Build, an agentic CLI for coding, building apps, and automating workflows is now available for SuperGrok Heavy subscribers. SuperGrok Heavy is normally USD 300/month but currently being offered at USD 99/month for 6 months.
[Grok Build Beta
Grok Build Beta adds a native subagent view, Plan Mode integration, mouse support, and a fullscreen terminal UI. Install with curl -fsSL https://x.ai/cli/install.sh | bash.
https://x.ai\
\
Saturday 16th May 2026
If you are in marketing/content creation, I think worth your while to dive a bit deeper into what Higgsfield just launched. I've just curated a few posts but I think in the coming days/weeks more will be coming out about this
=====
Introducing Higgsfield Supercomputer
The first ever cloud-native, self-learning AI agent for end-to-end task execution.
40+ built-in tools. Three layers of memory. Access via browser or Telegram.
Powered by enhanced Hermes Agent.
Introducing Higgsfield Supercomputer
The first ever cloud-native, self-learning AI agent for end-to-end task execution.
40+ built-in tools. Three layers of memory. Access via browser or Telegram.
Powered by enhanced Hermes Agent.
$39/month. That’s the current price of a content operation that routes GPT-5.5 Pro, Claude Opus, Seedance 2.0, Veo, and Kling through one agent and ships finished video, images, and copy 24/7.
Higgsfield Supercomputer just launched this. One prompt goes in. The agent decomposes
Introducing Higgsfield Supercomputer
The first ever cloud-native, self-learning AI agent for end-to-end task execution.
40+ built-in tools. Three layers of memory. Access via browser or Telegram.
Powered by enhanced Hermes Agent.
Higgsfield Supercomputer.
24/7 pocket marketing agency that gets smarter with every task.
> GPT 5.5, Opus 4.7, Kimi K2.6, Gemini. Freedom of choice across ecosystems.
> Plugs into Instagram, TikTok, YouTube, Slack.
> Access on mobile via Telegram.
> Picks up context from your team.
Introducing Higgsfield Supercomputer
The first ever cloud-native, self-learning AI agent for end-to-end task execution.
40+ built-in tools. Three layers of memory. Access via browser or Telegram.
Powered by enhanced Hermes Agent.
[Higgsfield Supercomputer - Agentic AI Content Creation
An agentic AI that runs your whole creative pipeline. Describe a reel, an ad, or a week of content - Higgsfield Supercomputer plans, generates, and delivers it all in one chat.
https://higgsfield.ai\
\
Elon Musk teasing at sizable capability upgrades in upcoming models from xAI
Grok 4.2 is based on foundation model V8, trained on Hoppers, with 0.5T params. Internally, xAI has completed a Grok V9 1.5T run, optimized for Nvidia Blackwell, and Elon says it is already looking great before Cursor data is added.
He says the difference between foundation model V8 and V9 is gigantic!
Our recently completed Grok V9 1.5T run is looking great and that is before Cursor data is added in supplemental training
The version numbers are a little confusing and deserve some explanation.
Internally, we are working on version 9 of our new foundation model, which is 1.5T params. This is substantially better in every way than v8: data curation, training recipe, size, etc. It is also optimized
Google just published a new resource for optimizing for generative AI in Google Search
====
In the new guide, optimizing your website for generative AI features on Google Search, you'll find:
Guidance on the importance of providing valuable, unique, non-commodity content
Tips about providing local, shopping, image, and video content
Mythbusting common "AEO/GEO" misconceptions
Initial guidance related to AI agents, a quickly emerging and evolving space
Information about why SEO best practices remain relevant and foundational to success with our generative AI features
[Google's Guide to Optimizing for Generative AI Features on Google Search
| Google Search Central
| Documentation
| Google for Developers
Learn how to optimize your website for Google Search's generative AI features, including official best practices, technical SEO advice, and emerging AI agent guidance.
https://developers.google.com\
\