Fernando Herrera

LatAm/Canadian-preneur: AI, consensus algos, cybersec, real estate & outdoors. Superpowered by a lovin family! Opinions r my own; RTs/follows R'not endorsements

[Fernando Herrera @ferherrera]

https://t.co/DJvs93p3f6

[Databricks @databricks]

IP Functions are now GA on Databricks!

Every firewall, load balancer, and application server generates records keyed on IP addresses. Analyzing that data in SQL used to mean regex hacks, custom UDFs, and brittle bitwise math.

Now you can parse, validate, and join IPv4 and IPv6 addresses and CIDR blocks with built-in SQL functions. They're optimized in Photon.

⚡ Up to 3.1x faster, 6.4x cheaper CIDR joins in benchmarks versus another leading cloud data warehouse.

IP Functions are a critical building block for our Security Lakehouse vision: threat detection, investigation, and network analytics on one governed copy of your data.

Learn more: https://t.co/XG6XaojKJz

[Fernando Herrera @ferherrera]

https://t.co/o24Q6h6Y1n

[Cern Basher @CernBasher]

https://t.co/KSdOBYepGt

[Farzad 🇺🇸 🇮🇷 @farzyness]

Easily one of the most impressive use-cases for agentic AI that I've come across thus far.

I asked my Grok @bot to make me a daily video on the latest happenings in frontier AI development around the tech, regulations, and development from the last 24 hours.

The below is its one-shot result.

It used DaVinci to edit the video using MCP with the assistance of Fable/Astra.

It automatically captured all information/clips/images/sources.

It automatically wrote the script and voiced it using @ElevenLabs with my trained voice.

It automatically edited and exported the file through DaVinci and dropped it in my downloads folder on my laptop.

Here's the full unedited prompt dictated through Whispr Flow:

I want to start a brand new project now. I don't know if we're going to need new agents for this but I'll let you decide.
What I want to do is create a daily video, primarily internal but also outsourced. We'll put it on a YouTube channel, my channel, or whatever, and turn it into a newsletter where I want to stay on top of all the breaking advancements in AI. I want to stay on top of not just the advancements but also the current events happening, all the leaders, and all the policy around it. I want to always be able to find the highest signal stuff using social cues, like videos that get a lot of views, posts that get a lot of views, or from very well-respected sources that understand the cutting edge.
I want to curate the absolute best places to find signal about what's going on in the space of AI. Then I want to turn that into a transcript that we will then turn into a video using our entire DaVinci process with MCP. Think of it as a hardcore news sort of format where the video just goes through the facts of what is happening and the things that are involved, like elite journalism to really describe what is happening.
For the video itself we would:
- draw sources and put them on the screen
- draw B-roll and put it on the screen
- add clips to it where relevant, snippets of a leader or somebody talking about those things
- add relevant imagery and all that stuff
It needs to be top-tier elite-looking, both from the look perspective and the content perspective. We can use Eleven Labs for the voice. You should already have things for that. Just use my voice, my voice profile that's there for the voice itself. The transcript we want to use: we don't want to use my voice. We want to use a very facts-based, neutral, journalistic, to-the-point, zero-opinion format: here's what's happening, here's how, here's why it's happening, here are the parties involved, etc.
For the visuals again, use the DaVinci sort of thing that we've set up on the Mac to put the video together. I would love an actual finished video today based on the latest that's happening. We want to focus on the last 24 hours.
Ideally the audience for this is people that are investors, high-intelligence individuals, and people that want to stay at the cutting edge but maybe don't have the time or the resources to do it. We want to become a primary place for these people to have access to this.
I would love to see what you can come up with and I will let you decide how to spin this up on the backend. We would want to do this literally daily and eventually want to have this automatically uploaded to a YouTube channel, either mine or another channel. We'll create clips out of it but ideally the piece of content itself needs to be so good. I think we have everything we need and it will be even sicker if we can literally bring in clips and add the clips. For example: "This is what happened today. This is all the things that are going on and blah blah blah." Let's say Elon Musk made a comment about this and we actually insert the clip of that comment and then we continue on, etc. Just freaking elite. I want this to be elite.

[Cern Basher @CernBasher]

https://t.co/r4JUwNlzqE

[Sarah Wang @sarahdingwang]

Ali Ghodsi explains how Databricks’ token consumption strategy has evolved:

“Around Q4 last year was when the models got really, really good, and we started noticing that it’s actually starting to give much better productivity.”

“It’s called Unity Gateway, this gateway used to provide token capacity. So you can get OpenAI, Anthropic, Gemini, Grok capacity.”

“Any customer can come to us, and we’ll just provide them that capacity because we have a relationship with those, and any open source model.”

“We started putting budget constraints in place and giving people warnings. Like, ‘You have this much of your budget left. You’re getting close to your ceiling.’ So we started doing that per person and per group, and then we started doing great analytics, so we could predict exactly where the costs were going.”

“We were able to actually bend the curve. The tokens have continued to go up, but the costs have been stagnant.”

Amazing chat @alighodsi @martin_casado

[Patrick Wendell @pwendell]

Today we rolled out Astra to every engineer at Databricks (N=~3500). Some notes that may be helpful to others:

1. Astra unambiguously out performs our previous highest-end models (Opus 5, Sol 5.6) on highly complex tasks, especially those related to high level system design or long range horizontal tasks.

2. Engineers given Astra increased overall coding spend by around 60% compared to baseline.

3. It is not clear Astra meaningfully improves on medium/low complexity coding tasks compared to earlier models. We suspect those tasks are mostly saturated (i.e. perfectly executed) by existing models.

4. We learned above by piloting Astra with around 200 users to gain signal on both quality and cost. We use Unity Gateway to do cohort-based experiments for all new models.

5. We give engineers a sub-budget specific to Astra to encourage them to use Astra selectively on complex tasks while preferring lower cost models for everyday tasks. Our engineers are able to mix-and-match tools and models within their overall budget envelope (we also allow for increased budgets through various mechanisms). These budgets are defined in Unity Gateway and regularly revisited.

Note: We do not have robust comparisons of Astra-vs-Fable because we have net yet rolled out Fable widely due to data retention policies.

[Marc Andreessen 🇺🇸 @pmarca]

https://t.co/lhP4mJq4Ad

[Omnigent @omnigent_ai]

From our recent Omnigent & Multi-Agent Coding Meetup in NYC, @dennylee (@databricks) gives a primer on why a meta-harness matters.

The talk covers:
🔹 Why coding agents and harnesses do not interoperate
🔹 How a meta-harness sits above Claude, Codex, Cursor, and similar tools
🔹 Multi-agent debates, where agents critique and improve each other’s code
🔹 Sharing a session so a teammate can view, comment, and run it if you grant them rights
🔹 Policies that require approval for file and shell operations

📽️ Watch now: https://t.co/G6WbGqXwKF

#Omnigent #AIAgents #NYC

[Nate Herk @nateherk]

https://t.co/tzTc5sUGbX

[Databricks @databricks]

GLM 5.3 is available as a day 0 release on Databricks!

GLM 5.3 is the smartest open-weight model available today and joins GLM 5.3 Flash and 30+ open-source and frontier models available natively on Databricks.

We host GLM 5.3 on secure GPUs owned by Databricks so you can use the most intelligent models with your most secure data.

With Unity Gateway, you can connect GLM 5.3 to your coding agents and get smart routing, centralized cost controls, and observability in a single pane of glass.

https://t.co/QhgYtiReY1

[Databricks @databricks]

Databricks Co-Founder and CEO @alighodsi has been named to the 2026 TIME100 AI list!

TIME highlights Ali in their AI leaders category saying, “as AI agents change how organizations work, Databricks is giving those agents access to secure, governed enterprise data on which to operate."

Congratulations to Ali and all of this year’s #TIME100AI honorees!

View the full list: https://t.co/ijcwk7Mo8g

[Omnigent @omnigent_ai]

Omnigent v0.11.0 is now available! 🎉

🎛️ Native harness control (Claude Code permission modes, Codex Max/Ultra)
⏱️ Automations: max_cost_usd + per-task permission mode
⌨️ Faster web and desktop flow (Cmd/Ctrl+N, Plan card)
🎨 Project emoji icons

🔗 Full details, bug fixes, and upgrade notes:
https://t.co/Otf1u5CCzw

#Omnigent #AIAgents

[cvxv666 @antpalkin]

Grok Bot by Elon Musk just replaced a $294,000 Wall Street research desk. Six AI agents, $200 a month, doing the work BlackRock pays people six figures for.

The guy who built it published every prompt.

What that desk cost before the bots:

> $27,000 a year for the Bloomberg terminal
> $22,000 for Refinitiv
> $50,000 for sell-side research
> $15,000 for AlphaSense
> $180,000 for the junior analyst who reads all of it until 2am so somebody senior can trade on it before the open

$294,000 a year. The agents cost $2,400. Same brief, 122 times cheaper, and it is there before the open.

Six agents split that desk. Each one gets its own computer in the cloud, and they all write into the same vault.

> FILINGS reads every 10-K, 10-Q and 8-K on a 100-ticker watchlist overnight and flags the ugly parts: going concern language, auditor changes, restatements
> EARNINGS reads the call transcript within 24 hours and tells you if the CFO got quieter than he was last quarter
> SECTOR does one pass per sector and picks up the rule change nobody read
> INSIDER catches Form 4 buys over $1 million and new positions from Two Sigma, Third Point and D.E. Shaw the day the 13Fs land
> CHATTER counts mentions on X and flags any ticker three standard deviations louder than its own 30-day normal
> CHIEF OF STAFF reads the other five at 5:30am, bins anything only one of them flagged, and emails you the ranked brief at 6

One analyst covers 30 names. Six agents cover 3,000 and finish before the open.

You show them the job once. They repeat it every night with your laptop shut.

Every research floor on Wall Street was priced on reading being slow and people being expensive. Both of those stopped being true this month.

So you wake up, read for five minutes, and know what moved on every name you hold. That used to be somebody's entire job, and he got a bonus for it.

The article below is the step by step guide to building the whole AI trading machine. Save & read it, you will want it open while you build.

[Ali Ghodsi @alighodsi]

This is true. It wasn't actually possible before 2010 because datacenter networks would bottleneck. We used to design coupled storage/compute where you brought compute close to "big data". But research on "full bisection bandwidth" networks made it possible to essentially just talk from any machine to the storage system at full speed. The disaggregation started then! Databricks and Snowflake started soon after many others followed. Now "Put it on the object store" is the way to go.

[Ali Ghodsi @alighodsi]

An extremely important functionality for agents is to simply extract fields out of PDFs. This turns out to be harder than people think because LLMs are primarily trained on predicting the next tokens. This leads them to "autocorrect" things that they shouldn't autocorrect. We launched an AI Extract capability that just excels at doing just this task with very high accuracy (95% vs 87% for others) and extremely low cost. Check out this blog on how we did it. The function can of course be called directly from SQL and be used throughout the platform.

https://t.co/60XC2mIZ2E

[SCOTTY BEAM @ScottyBeamIO]

WTF, THIS GUY FROM CURSOR GAVE GROK BOT A PILE OF CLOTHES HIS SISTER WAS TOO LAZY TO SELL – AND IT HANDLED THE ENTIRE SALE BY ITSELF

His sister had tons of clothes she never wore anymore, but selling them meant taking photos, creating listings, adding descriptions, answering buyers and negotiating prices – so she kept putting it off.

He gave the entire task to Grok Bot.

It looked at the photos, identified the items, created the listings, pulled in the relevant information and started negotiating with buyers on its own.

No babysitting. No approving every message. No sitting there telling it what to do next.

Grok Bot has its own computer, so once he handed over the task, it just kept going.

And that’s what makes this interesting.

The biggest problem with small tasks isn’t usually the work itself.

It’s the mental load of remembering them, starting them, checking them, replying to people and making sure they actually get finished.

AI agents are starting to remove that entire layer.

You don’t need to ask AI how to sell your clothes.

You can just say: “Sell these”. And let it figure out the rest.

Everyone has a list of things they’ve been putting off for months.

AI is starting to check them off for you.

[Fernando Herrera @ferherrera]

https://t.co/Ai4GUeLyZV

[Fernando Herrera @ferherrera]

https://t.co/OwRju8h4Xd