3 leanings from re-building COMPUTE for AI Agents.
The model will make mistakes. when it does, the sandbox shouldn’t hold anything that an attacker can use against it.
Agents spend most of their time waiting. An idle agent should cost almost nothing and wake up faster than a nt becomes more useful the more it remembers. Its workspace should survive between turns, between days, and between versions of the operating system.
The model will make mistakes. when it does, the sandbox shouldn’t hold anything that an attacker can use against it.
Agents spend most of their time waiting. An idle agent should cost almost nothing and wake up faster than a nt becomes more useful the more it remembers. Its workspace should survive between turns, between days, and between versions of the operating system.
35
12
83
This is how EACH task gets its own computer for Gemini Agent.
MicroVMs built for AI Agents that can rapidly go from 0 to millions of isolated computers.
Be REALLY AMBITIOUS with your Gemini Agent.
MicroVMs built for AI Agents that can rapidly go from 0 to millions of isolated computers.
Be REALLY AMBITIOUS with your Gemini Agent.
Gemini Agent is built for high-ceiling work where EACH task gets its own computer.
You can choose Gemini 4 Argon or Claude Opus 5.5.
10 things that make it a universal work Agent (read till the end to see what a REAL work agent looks like):
1. One Agent, one API
Chat, and code generation from a single interface. Assign it work, schedule it, or have it respond to events.
2. Coworker agents with their own identities
It creates job-specific sub-agents, each with its own identity, and computer running in parallel for hours or days.
Each Coworker agent gets its own Workspace account: an agent email, calendar, Drive, and presence in your company directory. You mention it in Chat. It acts under its own identity, not yours, and sees only what you share.
3. Everywhere, including headless
Web, iOS, Android, Windows, Mac, command line, Workspace, Microsoft 365, Slack. It also runs as a headless agent with no dedicated interface.
4. Work 24/7 in the cloud
It keeps one set of memories, context, and personalization graph across every device and channel. Work that takes hours or days keeps running after you close your laptop. You never have to re-brief it.
5. Layers of memory that improve over time
- Session: the task in front of it, even when it runs for days.
- Semantic: a structured knowledge base it builds as it reads documents, talks to people, and works with other agents.
- Procedural: how a job gets done, including skills it writes for itself.
- Episodic: everything it has done before.
6. Proactive by design
It does not wait to be prompted. An email asking you for a deck gets flagged as delegatable, one click to hand off. Your inbox ranks by what matters, not by what just arrived, and tells you why.
7. Tools and skills, with company registries
Salesforce, ServiceNow, Jira, Git, Confluence, Microsoft 365, Slack, BigQuery, Databricks, Postgres, Snowflake, your desktop files, and any MCP server inside or outside your network. Teams publish their own tools and skills to a shared company registry.
8. Model Choice so you are not locked in
It runs each job on the model that fits best, running across the Gemini and Claude models from Anthropic today, with other private and open models to come.
9. Governed like an employee
Every agent gets a cryptographically attested identity with least-privilege permissions, and an audit trail attributed to the agent rather than a person. Agents run in a sandbox behind Agent Gateway, an AI network firewall. Write the policy once and it applies to every agent.
10. Costs you can actually cap
Smart Routing puts each workload on the model with the best performance at the lowest cost. Set a hard spend limit in Cloud Billing and that project's agent pauses until you resume it in one click.
Thank you for reading so far. Stay tuned for updates.
We are 100% committed to make it the best Agent for work.
24
12
178
Google's Ambient quality agent runs 24/7 in LOOP to monitor and fix AI agents in production.
It reads live traces.
Catches silent failures.
Points to the exact line of code to fix.
And it's 100% open-source.
GitHub: github.com/google/adk-rec…
It reads live traces.
Catches silent failures.
Points to the exact line of code to fix.
And it's 100% open-source.
GitHub: github.com/google/adk-rec…
75
34
320
Super cool and meditative 🧘♂️
this is just ~56 KB HTML file
one shotted with gemini 4 argon using antigravity
this is just ~56 KB HTML file
one shotted with gemini 4 argon using antigravity
39
10
216
Excaliflow just got multi-canvas tabs and auto-save.
You asked for it on X, so I shipped it.
Added this with Gemini 4 Argon in Antigravity in one session.
It's 100% open-source: github.com/Shubhamsaboo/a…
You asked for it on X, so I shipped it.
Added this with Gemini 4 Argon in Antigravity in one session.
It's 100% open-source: github.com/Shubhamsaboo/a…
This is NOW 100% open-source.
Built with Gemini 4 Argon in Antigravity (plus gemini 4 argon is now available for pro users).
GitHub Repo: github.com/Shubhamsaboo/a…
24
17
258
I wanted an animated Excalidraw so asked Gemini 4 Argon to buid one.
And damm it really cooked. Fully working!!
Will make it 100% open-source.
And damm it really cooked. Fully working!!
Will make it 100% open-source.
46
32
519
Watch the full video on YouTube: youtube.com/watch?v=gRhM1h…
3
0
25
Google Gemini 4 Argon just went LIVE today.
10 INSANE things our team built with Gemini 4 Argon.
last one is my favourite.
10 INSANE things our team built with Gemini 4 Argon.
last one is my favourite.
24
12
148
@minchoi Here's the full video with top 10 use-cases: youtube.com/watch?v=gRhM1h…
0
1
7
This is INSANE.
Google just dropped a free AI notetaker app for Mac.
Every premium feature works without any subscription.
100% free and fully local, powered by Embedding Gemma 2.
Google just dropped a free AI notetaker app for Mac.
Every premium feature works without any subscription.
100% free and fully local, powered by Embedding Gemma 2.
268
472
7,366
Gemini Agent is built for high-ceiling work where EACH task gets its own computer.
You can choose Gemini 4 Argon or Claude Opus 5.5.
10 things that make it a universal work Agent (read till the end to see what a REAL work agent looks like):
1. One Agent, one API
Chat, and code generation from a single interface. Assign it work, schedule it, or have it respond to events.
2. Coworker agents with their own identities
It creates job-specific sub-agents, each with its own identity, and computer running in parallel for hours or days.
Each Coworker agent gets its own Workspace account: an agent email, calendar, Drive, and presence in your company directory. You mention it in Chat. It acts under its own identity, not yours, and sees only what you share.
3. Everywhere, including headless
Web, iOS, Android, Windows, Mac, command line, Workspace, Microsoft 365, Slack. It also runs as a headless agent with no dedicated interface.
4. Work 24/7 in the cloud
It keeps one set of memories, context, and personalization graph across every device and channel. Work that takes hours or days keeps running after you close your laptop. You never have to re-brief it.
5. Layers of memory that improve over time
- Session: the task in front of it, even when it runs for days.
- Semantic: a structured knowledge base it builds as it reads documents, talks to people, and works with other agents.
- Procedural: how a job gets done, including skills it writes for itself.
- Episodic: everything it has done before.
6. Proactive by design
It does not wait to be prompted. An email asking you for a deck gets flagged as delegatable, one click to hand off. Your inbox ranks by what matters, not by what just arrived, and tells you why.
7. Tools and skills, with company registries
Salesforce, ServiceNow, Jira, Git, Confluence, Microsoft 365, Slack, BigQuery, Databricks, Postgres, Snowflake, your desktop files, and any MCP server inside or outside your network. Teams publish their own tools and skills to a shared company registry.
8. Model Choice so you are not locked in
It runs each job on the model that fits best, running across the Gemini and Claude models from Anthropic today, with other private and open models to come.
9. Governed like an employee
Every agent gets a cryptographically attested identity with least-privilege permissions, and an audit trail attributed to the agent rather than a person. Agents run in a sandbox behind Agent Gateway, an AI network firewall. Write the policy once and it applies to every agent.
10. Costs you can actually cap
Smart Routing puts each workload on the model with the best performance at the lowest cost. Set a hard spend limit in Cloud Billing and that project's agent pauses until you resume it in one click.
Thank you for reading so far. Stay tuned for updates.
We are 100% committed to make it the best Agent for work.
You can choose Gemini 4 Argon or Claude Opus 5.5.
10 things that make it a universal work Agent (read till the end to see what a REAL work agent looks like):
1. One Agent, one API
Chat, and code generation from a single interface. Assign it work, schedule it, or have it respond to events.
2. Coworker agents with their own identities
It creates job-specific sub-agents, each with its own identity, and computer running in parallel for hours or days.
Each Coworker agent gets its own Workspace account: an agent email, calendar, Drive, and presence in your company directory. You mention it in Chat. It acts under its own identity, not yours, and sees only what you share.
3. Everywhere, including headless
Web, iOS, Android, Windows, Mac, command line, Workspace, Microsoft 365, Slack. It also runs as a headless agent with no dedicated interface.
4. Work 24/7 in the cloud
It keeps one set of memories, context, and personalization graph across every device and channel. Work that takes hours or days keeps running after you close your laptop. You never have to re-brief it.
5. Layers of memory that improve over time
- Session: the task in front of it, even when it runs for days.
- Semantic: a structured knowledge base it builds as it reads documents, talks to people, and works with other agents.
- Procedural: how a job gets done, including skills it writes for itself.
- Episodic: everything it has done before.
6. Proactive by design
It does not wait to be prompted. An email asking you for a deck gets flagged as delegatable, one click to hand off. Your inbox ranks by what matters, not by what just arrived, and tells you why.
7. Tools and skills, with company registries
Salesforce, ServiceNow, Jira, Git, Confluence, Microsoft 365, Slack, BigQuery, Databricks, Postgres, Snowflake, your desktop files, and any MCP server inside or outside your network. Teams publish their own tools and skills to a shared company registry.
8. Model Choice so you are not locked in
It runs each job on the model that fits best, running across the Gemini and Claude models from Anthropic today, with other private and open models to come.
9. Governed like an employee
Every agent gets a cryptographically attested identity with least-privilege permissions, and an audit trail attributed to the agent rather than a person. Agents run in a sandbox behind Agent Gateway, an AI network firewall. Write the policy once and it applies to every agent.
10. Costs you can actually cap
Smart Routing puts each workload on the model with the best performance at the lowest cost. Set a hard spend limit in Cloud Billing and that project's agent pauses until you resume it in one click.
Thank you for reading so far. Stay tuned for updates.
We are 100% committed to make it the best Agent for work.
172
99
859
Markdown is now native in Google Docs and Drive.
Hermes, Grokbot, Muse, and Instinct can share one living source of truth: a file I can open, read, and change anytime.
The agents write to it. I edit it. The next agent picks up where I left off.
Hermes, Grokbot, Muse, and Instinct can share one living source of truth: a file I can open, read, and change anytime.
The agents write to it. I edit it. The next agent picks up where I left off.
30
8
167
Watch the full 30 mins masterclass for free: youtube.com/watch?v=iNLeFD…
0
3
16
Just SHIPPED: a free 30 mins masterclass to build a multimodal voice AI Agent that can see, think, talk, and draw at the same time.
Go from an problem → idea → spec → working agent as an AI PM.
Go from an problem → idea → spec → working agent as an AI PM.
11
11
67
Introducing Google Cloud Advent of Agents Season 3.
31 days of building secure, governed AI agents for production.
Daily tutorials. Copy-paste code. 100% free.
Starts tomorrow: adventofagents.com
31 days of building secure, governed AI agents for production.
Daily tutorials. Copy-paste code. 100% free.
Starts tomorrow: adventofagents.com
32
31
184
Crazyy..we got AI Agents talking to each other before GTA-6.
GrokBot can ask Muse to make a call, muse can tell grok how it went. Bring it Instinct and Claude Code to the mix.
And it's 100% opensource.
GrokBot can ask Muse to make a call, muse can tell grok how it went. Bring it Instinct and Claude Code to the mix.
And it's 100% opensource.
26
12
79
INSANE. Voice AI Agents getting better every other week.
I'm now talking to a live avatar of my insurance claim voice agent. It can see, talk, think and draw in real-time now with a face.
Switched to Hindi mid-call and it still worked.
Made it 100% open-source.
I'm now talking to a live avatar of my insurance claim voice agent. It can see, talk, think and draw in real-time now with a face.
Switched to Hindi mid-call and it still worked.
Made it 100% open-source.
46
28
352
Crazy...my wealth advisor agent made 12 compliance violations in a single reply.
Then I added ZeroDrift Anchor scanner. It caught the risky lines, cited the exact rules, rewrote the response, and blocked it when it still failed.
This is HOW AI agents should SHIP in 2026.
fandf.co/3VJf5vi
Then I added ZeroDrift Anchor scanner. It caught the risky lines, cited the exact rules, rewrote the response, and blocked it when it still failed.
This is HOW AI agents should SHIP in 2026.
fandf.co/3VJf5vi
Introducing Anchor 3.0 by @zerodrift_ai, the first enforcement model built to secure AI agents and keep everything they say compliant at runtime.
It caught more compliance violations than every frontier model we tested, up to 34x faster and up to 12x cheaper.
Live today, first ://t.co/oyagIt4dn1
12
5
37
Introducing a new way to EDIT with @typesafeai Jev and gemini flash lite.
Change one thing and it finds everything else that needs to change (with suggested fixes). Super fast and near real-time.
Will make it 100% open-source Chrome extension.
Change one thing and it finds everything else that needs to change (with suggested fixes). Super fast and near real-time.
Will make it 100% open-source Chrome extension.
23
15
244
Introducing a new way FIND (⌘F) with @typesafeai Jev.
Find what you mean, not what just matches. Super fast and near real-time.
Made it 100% open-source chrome extension.
Find what you mean, not what just matches. Super fast and near real-time.
Made it 100% open-source chrome extension.
78
104
2,312
Google just published a paper on recursive self-improvement.
Dream RSI. An agent MUST dream to recursively self-improve. History is the world it dreams in.
Dream RSI. An agent MUST dream to recursively self-improve. History is the world it dreams in.
20
37
247
Holy moly...this agent skill tells you where readers stop reading your draft.
It's not another no-ai slop skill - it actually reads your draft like a human reviewer and drop precise comments.
comes with built-in UI to make things visual.
As always...this is 100% open-source.
It's not another no-ai slop skill - it actually reads your draft like a human reviewer and drop precise comments.
comes with built-in UI to make things visual.
As always...this is 100% open-source.
7
6
59
My mind is f*** blown.
I built this live insurance claim agent that can see, talk, think and draw in real-time using the new Gemini 3.8 LIVE.
Even switched my language to Hindi mid-call and it still worked.
VOICE AI can't be more real. Made it 100% open-source.
I built this live insurance claim agent that can see, talk, think and draw in real-time using the new Gemini 3.8 LIVE.
Even switched my language to Hindi mid-call and it still worked.
VOICE AI can't be more real. Made it 100% open-source.
80
90
1,359
5 INSANE things to do with Grok Bot right NOW
- Give it a phone number and inbox
- Screenshot anything and text your Chief of Staff
- last30days for full sweep on any topic
- Auto sync login from chrome to grok bot
- Install Claude Code on its computer
youtube.com/watch?v=ZoYDq5…
- Give it a phone number and inbox
- Screenshot anything and text your Chief of Staff
- last30days for full sweep on any topic
- Auto sync login from chrome to grok bot
- Install Claude Code on its computer
youtube.com/watch?v=ZoYDq5…
9
10
79
Crazyy that this even works.
Agent Reach gives your Hermes and OpenClaw agents access to X, LinkedIn, Facebook, Instagram and more.
And it's 100% free and Opensource.
Agent Reach gives your Hermes and OpenClaw agents access to X, LinkedIn, Facebook, Instagram and more.
And it's 100% free and Opensource.
25
55
564
Watch the full video on YouTube, and if you find it useful, please drop a like and share it with your friends.
youtube.com/watch?v=Q08HKd…
youtube.com/watch?v=Q08HKd…
0
2
19
I just shipped a free crash course on building AI agents without writing a single line of code.
In 20 mins, I take one problem through the full agent development lifecycle using just a coding agent.
You can use Codex, Claude Code, Antigravity or anything else.
In 20 mins, I take one problem through the full agent development lifecycle using just a coding agent.
You can use Codex, Claude Code, Antigravity or anything else.
25
20
134
JUST SHIPPED: 8 free Agent skills with evals. 100% free and open-source.
Here are my top 3 picks:
/advisor-orchestrator-worker
Your agent becomes the orchestrator of a three-tier model team. GPT-6 runs the orchestration, Gemini 3.8 does the cheap parallel worker tasks, and onsulted as the advisor.
/first-reader
Simulates real readers going through your draft and reports where they lose interest, where they stop reading, and what they remember afterward, without rewriting a word.
/project-graveyard: It scans your machine, works out why each one died from its git history, shows you your own death patterns, and resurrects the one with a pulse.
🌟 agent skills github repo: github.com/Shubhamsaboo/a…
Here are my top 3 picks:
/advisor-orchestrator-worker
Your agent becomes the orchestrator of a three-tier model team. GPT-6 runs the orchestration, Gemini 3.8 does the cheap parallel worker tasks, and onsulted as the advisor.
/first-reader
Simulates real readers going through your draft and reports where they lose interest, where they stop reading, and what they remember afterward, without rewriting a word.
/project-graveyard: It scans your machine, works out why each one died from its git history, shows you your own death patterns, and resurrects the one with a pulse.
🌟 agent skills github repo: github.com/Shubhamsaboo/a…
20
24
158
I just sat down with @mvanhorn to talk about all things Grok @bot.
DON'T MISS OUT, it's a free masterclass.
00:00 Intro
01:22 Who Matt Van Horn is
02:44 Why Grok Bot is the first agent built for humans
04:32 The tweet, "cap bro," and Elon's repost
07:01 The 1.2M-view hacks t-all shortcut
08:56 Installing Grok Bot from scratch
09:55 The real killer feature: Mac and iOS sync
10:41 Setting up your chief of staff
11:20 The Google Apps CLI that makes Sheets instant
12:21 Screenshot to calendar, and the daily family email
13:13 The bot that reset its own password
14:25 Dr. Eggbot, the bot that builds bots
15:03 Building an ELI5 bot in one prompt
16:47 The soccer tracker that corrected itself
18:35 What compound engineering actually is
20:05 CE plan, CE work, and never reading the plan
21:39 Using CE plan to DM hundreds of people
23:47 Agent Mail, and why a bot needs its own inbox
25:45 Giving a bot a phone number
27:12 Booking lunch in Portugal, and Pedro hanging up
29:01 The best voice setup for talking to your agent
30:58 Building a visual ELI5 bot live on the call
32:28 Last 30 Days, the second most popular bot
34:29 Pulling Reddit, X and Polymarket signal on one topic
37:21 The phone call finally works
40:14 Why the cron job is the actual magic
41:54 Silence is the hack
42:44 Agent Cookie and Tailscale
44:37 Printing CLIs, and why clicking burns tokens
45:33 Matt's top three workflows
47:36 Running Fable 5 inside Grok Bot
DON'T MISS OUT, it's a free masterclass.
00:00 Intro
01:22 Who Matt Van Horn is
02:44 Why Grok Bot is the first agent built for humans
04:32 The tweet, "cap bro," and Elon's repost
07:01 The 1.2M-view hacks t-all shortcut
08:56 Installing Grok Bot from scratch
09:55 The real killer feature: Mac and iOS sync
10:41 Setting up your chief of staff
11:20 The Google Apps CLI that makes Sheets instant
12:21 Screenshot to calendar, and the daily family email
13:13 The bot that reset its own password
14:25 Dr. Eggbot, the bot that builds bots
15:03 Building an ELI5 bot in one prompt
16:47 The soccer tracker that corrected itself
18:35 What compound engineering actually is
20:05 CE plan, CE work, and never reading the plan
21:39 Using CE plan to DM hundreds of people
23:47 Agent Mail, and why a bot needs its own inbox
25:45 Giving a bot a phone number
27:12 Booking lunch in Portugal, and Pedro hanging up
29:01 The best voice setup for talking to your agent
30:58 Building a visual ELI5 bot live on the call
32:28 Last 30 Days, the second most popular bot
34:29 Pulling Reddit, X and Polymarket signal on one topic
37:21 The phone call finally works
40:14 Why the cron job is the actual magic
41:54 Silence is the hack
42:44 Agent Cookie and Tailscale
44:37 Printing CLIs, and why clicking burns tokens
45:33 Matt's top three workflows
47:36 Running Fable 5 inside Grok Bot
7
7
62
I'm open-sourcing /first-reader, a free agent skill that reads your draft the way an actual person would.
first-reader simulates two people going through your draft one passage at a time, and reports how they felt - what's interesting, what doesn't make sense, things they thing they will remember the next day.
A third reader just skims and tells you whether they'd have opened it at all.
You point your agent at a draft and say who it's for:
"review this, the audience is backend engineers who've seen ten launches this month"
A few minutes later you get the read back in plain language, plus a page with your draft on it and each reader's comments sitting next to the lines they reacted to. You can see where they leaned in, where they bailed, and a strip down the side tracks their attention rising and falling as they go.
Get the free, open-source skill here: github.com/Shubhamsaboo/a…
If you find it useful, please ⭐ the repo and share it with anyone interested in writing better and clearer.
first-reader simulates two people going through your draft one passage at a time, and reports how they felt - what's interesting, what doesn't make sense, things they thing they will remember the next day.
A third reader just skims and tells you whether they'd have opened it at all.
You point your agent at a draft and say who it's for:
"review this, the audience is backend engineers who've seen ten launches this month"
A few minutes later you get the read back in plain language, plus a page with your draft on it and each reader's comments sitting next to the lines they reacted to. You can see where they leaned in, where they bailed, and a strip down the side tracks their attention rising and falling as they go.
Get the free, open-source skill here: github.com/Shubhamsaboo/a…
If you find it useful, please ⭐ the repo and share it with anyone interested in writing better and clearer.
16
26
162
Google engineers dropped security review skills for AI coding agents.
It can find, reproduce and patch vulnerabilities autonomously.
100% Opensource.
It can find, reproduce and patch vulnerabilities autonomously.
100% Opensource.
25
60
437
GPT-6 Astra keeps giving.
It's WILD that it can create an entire game from a prompt and few images from my old video game.
I gave it a /goal to make a fully-playable Road Rash game that I can play on my MacBook.
It ran overnight (~8 hrs) and I woke to this. INSANE.
It's WILD that it can create an entire game from a prompt and few images from my old video game.
I gave it a /goal to make a fully-playable Road Rash game that I can play on my MacBook.
It ran overnight (~8 hrs) and I woke to this. INSANE.
15
7
54
Anyone, LITERALLY ANYONE can now build generational wealth with AI Agents.
"Grok @bot feels like first Agent experience designed for humans."
I spent an hour with Matt Van Horn to talk about ALL THINGS Grok Bot. Every single workflow.
Dropping soon, 100% free.
"Grok @bot feels like first Agent experience designed for humans."
I spent an hour with Matt Van Horn to talk about ALL THINGS Grok Bot. Every single workflow.
Dropping soon, 100% free.
13
11
72
GPT-6 Astra is absolutely INSANE.
It created the old CONTRA in zero-shot in 30 mins.
Just like I remember from childhood. So COOL.
It created the old CONTRA in zero-shot in 30 mins.
Just like I remember from childhood. So COOL.
45
39
436
@bot Just recorded a super cool podcast with @mvanhorn on All things Grok bot (it was so much fun). Coming out soon.
x.com/mvanhorn/statu…
x.com/mvanhorn/statu…
1
2
10
dr Egg @bot creates super high-quality employees on autopilot.
at this point I am not even doing anything.
my chief of staff bot find recurring workflows and communicates with eggbot to create bots for them.
this is literally like living in the FUTURE.
at this point I am not even doing anything.
my chief of staff bot find recurring workflows and communicates with eggbot to create bots for them.
this is literally like living in the FUTURE.
10
5
23
Agentic video understanding with Gemini 3.7 Flash is HUGE.
Just reduced cost, less tokens, and higher accuracy. No biggie.
Just reduced cost, less tokens, and higher accuracy. No biggie.
7
38
903
MUST READ paper from Google Research.
Interesting bits...
1. Agents get better at tasks by keeping a wiki of everything they've learned.
2. Small model with skills beats one 3x its size without them.
3. Delete the wiki and the gains go away.
Interesting bits...
1. Agents get better at tasks by keeping a wiki of everything they've learned.
2. Small model with skills beats one 3x its size without them.
3. Delete the wiki and the gains go away.
35
45
407
How i run a self-improving Hermes agent organization built on four pillars.
• Direction: I set goals, approve, and correct when needed.
• Execution: Hermes chief-of-staff, specialists agents own domains, and CLIs act.
• Learning: Skills preserve what works. Memory stores omes and metrics improve the next run.
Goal is not just to complete more tasks. It is to build an agent org that gets smarter every day.
• Direction: I set goals, approve, and correct when needed.
• Execution: Hermes chief-of-staff, specialists agents own domains, and CLIs act.
• Learning: Skills preserve what works. Memory stores omes and metrics improve the next run.
Goal is not just to complete more tasks. It is to build an agent org that gets smarter every day.
20
18
132
Build your OWN harness and use ANY model you like.
Google ADK gives you cross-session memory, per-user sandbox, sub-agents, guardrails, and a nightly pass where the agent dreams.
100% Opensource.
Google ADK gives you cross-session memory, per-user sandbox, sub-agents, guardrails, and a nightly pass where the agent dreams.
100% Opensource.
We’re ending our partnership with Cursor following its acquisition by SpaceX. Under our proposal, Cursor’s direct access to our models would end on November 12.
We know that the people most affected by this decision are the developers who rely on OpenAI models in Cursor. We care about their experience in this transition and we’re ready to go above and beyond to support them.
openai.com/index/our-deci…
9
10
52
Crazyy...someone measured Claude's vocabulary across 47,464 GitHub pull requests.
A word cluster that didn't exist in 2025 is now 45% of human-authored PRs.
The top word is "load-bearing"
A word cluster that didn't exist in 2025 is now 45% of human-authored PRs.
The top word is "load-bearing"
78
99
1,654
So COOL...someone just built THE OFFICE theme agent harness.
Claude Code, Codex, Antigravity, and Cursor working with each other on your existing subscription limits.
100% Opensource.
Claude Code, Codex, Antigravity, and Cursor working with each other on your existing subscription limits.
100% Opensource.
25
27
489
