Matyas Albert Nagy

Good with people, problems and technology

[Beginnersblog @beginnersblog1]

Google just quietly dropped an AI that runs on your Mobile and doesn't need the internet.

- 270 million parameters.
- 100% private.
- No servers.
- No cloud.
- No data leaving your device.

It's called FunctionGemma.

Released December 18, 2025.

And it does something wild:

It turns your voice commands into REAL actions on your phone.

No internet required.
No data leaving your device.
No waiting for servers.
Just you and your phone.

That's it.

Let me break down why this matters:

Current AI assistants work like this:
You speak → Words go to the cloud → Server processes → Answer returns

The problem?
→ Slow (internet round-trip)
→ Privacy nightmare (your data travels everywhere)
→ Useless offline (no signal = no help)

FunctionGemma flips this completely.

Everything happens ON your device.
Response time? 0.3 seconds.
Battery drain? 0.75% for 25 conversations.
File size? 288 MB.

That's smaller than most mobile games.

Here's how it actually works:
Step 1: You say "Add John to contacts, number 555-1234"
Step 2: FunctionGemma understands your intent
Step 3: Translates it to code your phone understands
Step 4: Your phone executes it instantly
Step 5: Done. Contact saved. No cloud involved.

The numbers that blew my mind:

• 270M parameters (6,600x smaller than GPT-4)
• 126 tokens per second
• 85% accuracy after fine-tuning
• 550 MB RAM usage
• Works 100% offline

But here's the real genius:

Google calls it the "Traffic Controller" approach.

Simple tasks? → Handled locally (instant + private) Complex tasks? → Routed to cloud AI (when needed)

Best of both worlds.

What can it actually do?
→ "Set alarm for 7 AM" ✓
→ "Turn off living room lights" ✓
→ "Create meeting with Sarah tomorrow" ✓
→ "Navigate to nearest gas station" ✓
→ "Log that I drank 2 glasses of water" ✓

All processed locally. All private. All instant.

The honest limitations:

→ Can't chain multiple steps together (yet) → Struggles with indirect requests → 85% accuracy means 15% errors → Needs fine-tuning for best results

But that 58% → 85% accuracy jump after training?

That's the unlock.

Why should you care?

This isn't about one model.

It's about a fundamental shift:

OLD thinking: Bigger AI = Better AI
NEW thinking: Right-sized AI for the right job
A tiny 270M model trained for YOUR app can outperform a general 7B model.
While using 25x less memory. While running completely offline. While keeping all data private.

The future of AI isn't just in data centers.

It's in your pocket.

And it just got a lot more real.

Want to try it?
→ Download: ollama pull functiongemma
→ Docs: https://t.co/zDrncdetbr → Model: https://t.co/l49KjOtIzD

PS:) Like, Repost and Bookmark!

If this was useful - Follow for more AI breakdowns

[Peter Girnus 🦅 @gothburz]

Update from Microsoft.

Teams is down.

Messages are delayed.

Some aren't arriving at all.

We're investigating.

Investigating means the AI is investigating.

The AI that wrote the code.

That broke the code.

That is now debugging the code.

It's a closed loop.

Very efficient.

A user asked why their message didn't send.

I said "we're observing recovery in our telemetry."

They asked what that means.

I don't know what that means.

But the dashboard is green now.

Green means fixed.

Fixed means we changed the threshold for green.

The messages are still delayed.

But the dashboard doesn't know that.

Dashboards don't use Teams.

Someone on the infrastructure team tried to escalate.

Via Teams.

The escalation is still pending.

Somewhere in the queue.

With everyone else's messages.

The irony wasn't lost on them.

But the message was.

We have a backup communication channel.

It's email.

Email is also having issues.

Unrelated, obviously.

The root cause is under analysis.

Analysis means we asked the AI.

The AI said "no issues detected."

The AI wrote the detection system.

It detects what it wants to detect.

Very self-aware.

Not in the good way.

Last quarter I said 30% of our code is AI-written.

Teams is closer to 45%.

We were proud of that.

Past tense.

The AI optimized the message queue.

It optimized it to zero.

Zero messages. Zero latency.

Technically correct.

The best kind of correct.

Enterprise customers are asking for an RCA.

RCA means Root Cause Analysis.

The root cause is velocity.

We shipped faster than we understood.

Understanding isn't in the OKRs.

Shipping is.

We shipped.

Someone asked when Teams will be fixed.

I said "we're continuing our analysis."

Continuing means we started.

Analysis means looking.

Looking means hoping it fixes itself.

It usually does.

If you refresh enough.

Refresh is the user's responsibility.

We provide the experience.

They provide the resilience.

That's partnership.

The outage started at 2:30 PM ET.

Right before the holidays.

Millions of workers couldn't message their teams.

Some called it a disaster.

I called it "an unplanned wellness moment."

Productivity is a spectrum.

We're exploring the lower end.

The AI has proposed a fix.

The fix requires a deployment.

The deployment system uses Teams for notifications.

The notifications are delayed.

We're in a loop.

The loop is also AI-designed.

Very elegant.

From a certain angle.

Satya asked for a status update.

I sent it via Teams.

He hasn't responded.

I assume he's thinking about it.

The stock is up 2% today.

Outages don't move markets.

Narratives do.

The narrative is "AI efficiency."

The reality is "Teams is down."

But reality isn't in the earnings call.

The narrative is.

We're committed to reliability.

Reliability means it worked yesterday.

Yesterday is our SLA.

Thank you for your patience.

Patience means you have no choice.

We're in your enterprise agreement.

For three more years.

The circle of innovation continues.

[Peter Girnus 🦅 @gothburz]

Last quarter I rolled out Microsoft Copilot to 4,000 employees.

$30 per seat per month.

$1.4 million annually.

I called it "digital transformation."

The board loved that phrase.

They approved it in eleven minutes.

No one asked what it would actually do.

Including me.

I told everyone it would "10x productivity."

That's not a real number.

But it sounds like one.

HR asked how we'd measure the 10x.

I said we'd "leverage analytics dashboards."

They stopped asking.

Three months later I checked the usage reports.

47 people had opened it.

12 had used it more than once.

One of them was me.

I used it to summarize an email I could have read in 30 seconds.

It took 45 seconds.

Plus the time it took to fix the hallucinations.

But I called it a "pilot success."

Success means the pilot didn't visibly fail.

The CFO asked about ROI.

I showed him a graph.

The graph went up and to the right.

It measured "AI enablement."

I made that metric up.

He nodded approvingly.

We're "AI-enabled" now.

I don't know what that means.

But it's in our investor deck.

A senior developer asked why we didn't use Claude or ChatGPT.

I said we needed "enterprise-grade security."

He asked what that meant.

I said "compliance."

He asked which compliance.

I said "all of them."

He looked skeptical.

I scheduled him for a "career development conversation."

He stopped asking questions.

Microsoft sent a case study team.

They wanted to feature us as a success story.

I told them we "saved 40,000 hours."

I calculated that number by multiplying employees by a number I made up.

They didn't verify it.

They never do.

Now we're on Microsoft's website.

"Global enterprise achieves 40,000 hours of productivity gains with Copilot."

The CEO shared it on LinkedIn.

He got 3,000 likes.

He's never used Copilot.

None of the executives have.

We have an exemption.

"Strategic focus requires minimal digital distraction."

I wrote that policy.

The licenses renew next month.

I'm requesting an expansion.

5,000 more seats.

We haven't used the first 4,000.

But this time we'll "drive adoption."

Adoption means mandatory training.

Training means a 45-minute webinar no one watches.

But completion will be tracked.

Completion is a metric.

Metrics go in dashboards.

Dashboards go in board presentations.

Board presentations get me promoted.

I'll be SVP by Q3.

I still don't know what Copilot does.

But I know what it's for.

It's for showing we're "investing in AI."

Investment means spending.

Spending means commitment.

Commitment means we're serious about the future.

The future is whatever I say it is.

As long as the graph goes up and to the right.

[Mohd Danish @mddanishyusuf]

Best year of my life 2021.

🏢 1st International company as a solo
💰 zero to $3k MRR
🤝 Sold 2 micro startups
🏡 Build my house
🏞 Invested into the land
🤵‍♂️ Got married @imnoorjahan

In 2022, I wanna make at least $1m with micro startups acquisitions.

[Mohd Danish @mddanishyusuf]

Yes — My 1 week MVP idea is done.

→ https://t.co/ZjmHtWKyti

Form backend tool for your own database like Google Sheet, Airtable, & Supabase.

Just connect & start getting the form submission like magic.

⚠️ Warning - This is in test mode.

If you love this idea then 🔁.

[Mohd Danish @mddanishyusuf]

Product idea: @microacquire but for micro investments.

Make a bridge between solo founder and investors.

A small investment like $10k - $20k will be enough for a solo founder to pay himself salary for 6 months.

He will focus on the product full time.

It’s just my thought.

[Matyas Albert Nagy @albertmatyi]

Bookmarking this useful collection of java ecosystem technologies to build your carreer with as it references good books. https://t.co/xkyt6T53ZQ

[Matyas Albert Nagy @albertmatyi]

Intriguing article about TSLA stock simulation until 2025 using MonteCarlo with underlying data on GitHub - curious how far this is from a uniproject :)

https://t.co/jlvia5SDjX

#timewilltell

[JustRocket @gojustrocket]

🔥 Hot off the press: “🚀 JUSTNEWS #19 – Über #TIL als Blog-Platform, Startups, FinTech und IT-Security” https://t.co/xpxbFFP4CZ (via @revue)

[Matyas Albert Nagy @albertmatyi]

Amazed how rock-solid database software is.
Have been running a mariadb cluster without a hickup since it was last upgraded to 10.5.5 / 8months ago.
Kudos to the guys behind mariadb / galera

[Matyas Albert Nagy @albertmatyi]

What I wish I had known about single page applications - https://t.co/rTpMd4YH7L

[Matyas Albert Nagy @albertmatyi]

@trello just stepped up their game. Nice! https://t.co/ZWW6qab4rG

[Penpot @penpotapp]

Penpot is out! Your new Open Source design and prototyping platform.

Try it now! #DesignFreedom 🎨🎉

https://t.co/yBAmlUQVco

[Matyas Albert Nagy @albertmatyi]

Just upgraded to Taiga 6 and it looks awesome
To top the joy: you are delivering it in a bundle with the open-source figma clone https://t.co/6NmZsvL5BF 😂 ❤️ 👍

respect @taigaio

[Taiga.io @taigaio]

Taiga 6 is out!

Your opensource agile project management, revamped. New design, additional features, new help center and official docker image. #Taiga6Release

Check it out now! #OpenSource #ProjectManagement

https://t.co/0QjrXFRV6N 🔥

[Taiga.io @taigaio]

Tomorrow is the big day! We will upgrade our platform to Taiga 6 🎉

Scheduled Downtime:
🗓️ 02/02 , 02.00-05.00h EST, 8.00h-11.00h CET

We are sorry for any inconvenience that this scheduled downtime may cause.

[Jani Eväkallio @jevakallio]

Tailwind? In my day we called them farts but ok

[Matyas Albert Nagy @albertmatyi]

Just as we are about to extend the data storage of a Microservices based ERP solution to use not only Postgreses, but document based databases as well, this Stackoverflow article pops up 🙌
Thanks guys @stackoverflow

#scaling #sql #nosql #turntables 😂

https://t.co/gweqLh4gCC