|
|
Hi there, this is your daily ☕️ Techpresso.
|
|
|
In today's Techpresso:
|
|
💥 Elon Musk pledges to rebuild xAI 🥑 Meta delays AI model over performance concerns 👋 Adobe CEO Shantanu Narayen steps down after 18 years 🧠 China approves first-ever commercial brain implant 🍎 Apple cuts App Store fees in China to 25% 🎁 + 10 other news & articles you might like 🧰 + 5 trending tools 📚 + 3 trending papers & reports
|
|
FROM OUR PARTNER
The largest layoff of 2026 just hit: Jack Dorsey announced the elimination of nearly 40% of Block's workforce which is 4,000 people who didn't see it coming. And yet, most people are still convinced their job is safe. Outskill is the world's first AI learning platform where over 10+ Million Learners have learnt from top industry leaders like Microsoft, NVIDIA and Google. They are hosting a 2 day LIVE AI Mastermind where you'll build automations and create personalized agents. START LEARNING NOW!
|
💥 Elon Musk pledges to rebuild xAI
LINK
|
- Elon Musk publicly apologized for past hiring mistakes at xAI and said the company is reviewing old interview records to reach back out to promising candidates who were previously turned down.
- Musk compared the overhaul to Tesla's early days, writing that xAI "was not built right first time around" and is now being rebuilt from the foundations up as a full reset.
- Musk also revealed a formal collaboration between Tesla and xAI called "Macrohard" or "Digital Optimus," which aims to build AI systems that can perform the functions of entire companies.
|
🥑 Meta delays AI model over performance concerns
LINK
|
- Meta has pushed back the release of its new AI model, code-named Avocado, from March to at least May because it fell short of leading rivals on internal tests for reasoning, coding, and writing.
- Avocado outperformed Meta's previous model and Google's Gemini 2.5 from March but did not match Gemini 3.0 from November, and leaders discussed temporarily licensing Gemini to power Meta's AI products.
- Meta invested $14.3 billion in Scale AI last year and made its CEO, Alexandr Wang, chief AI officer, who built an internal lab called TBD Lab with around 100 employees working on Avocado.
|
👋 Adobe CEO Shantanu Narayen steps down after 18 years
LINK
|
- Adobe CEO Shantanu Narayen is stepping down after nearly 18 years leading the company and will stay on as board chair while Adobe searches for his successor.
- Under Narayen, Adobe grew from under $1 billion in revenue to over $25 billion, but generative AI tools now challenge its core creative software business.
- The shift comes as tech firms cut thousands of jobs to reorganize around AI, with companies like Atlassian and Block recently eliminating roughly 5,600 positions combined.
|
🧠 China approves first-ever commercial brain implant
LINK
|
- China has approved the first-ever commercial brain implant, a brain-computer interface made by Neuracle Medical Technology, for use in people with spinal cord injuries.
- The coin-sized wireless device sits on the brain's surface and records electrical signals from neurons, which software decodes to let patients control things like a computer cursor.
- No BCI devices have been approved for commercial use in the U.S., where Neuralink, Synchron, and Paradromics are still running clinical trials with their own implants.
|
🍎 Apple cuts App Store fees in China to 25%
LINK
|
- Apple is dropping its standard App Store commission rate in China from 30 percent to 25 percent for in-app purchases on iOS and iPadOS, starting March 15, after discussions with Chinese regulators.
- Developers in the App Store Small Business or Mini Apps programmes will also see their fees cut by 3 percent, going from 15 to 12 percent on IAPs and subscription renewals after the first year.
- Developers don't need to agree to the new terms by the March 15 deadline to receive the lower rates, and the change comes a year after reports of a state watchdog investigating Apple's fees.
|
|
FROM OUR PARTNER
🗣️Voice Agents need speed you trust Speechmatics' Voice Agent API delivers partials in <250ms and finals around 250-300ms. Built real-time first, not batch models forced into streaming. That's why your voice agents get unmatched accuracy at conversational latency. What you get: - 55+ languages, accent-agnostic performance
- Native LiveKit, Pipecat, Vapi support
- Deployment in the cloud, on-prem, hybrid, on-device
👉 Start building with $200 free credits. |
|
|
|
Other
news & articles you might like
-
Claude can now create interactive charts and visualizations directly in chat
LINK
-
Google Chrome on Linux is getting a big upgrade
LINK
-
Facebook Marketplace’s new AI ends the “is this available” nightmare
LINK
-
Lucid Motors shows off robotaxi concept called ‘Lunar’
LINK
-
Bumble introduces an AI dating assistant, ‘Bee’
LINK
-
China's ByteDance Outsmarts US Sanctions With Offshore Nvidia AI Buildout
LINK
-
Meta's massive undersea cable project delayed in Persian Gulf as Iran conflict escalates
LINK
|
|
🧰 Trending tools
|
Pinnacle: measures brain activity via iPhone sensors and provides conversational AI coaching to improve focus, resilience, and mental performance through real-time feedback.
LINK
|
|
GStack: turns Claude into specialized workflow assistants with slash commands for code reviews, deployment, browser automation, and engineering retrospectives to streamline development tasks.
LINK
|
|
Pre: an AI accountability agent that tracks startup progress through connected data and sends unfiltered weekly reports to keep founders focused on goals.
LINK
|
|
Manus Agents for Telegram: runs multi-step AI tasks, research, and file operations directly in Telegram via QR code connection without terminal or API setup.
LINK
|
|
Algorithms and Data Structures in TypeScript: a free 400-page book covering sorting, dynamic programming, graphs, trees, heaps, and hash tables with executable, tested code examples.
LINK
|
|
You can check the previous tools here, or add your tool here
|
|
📚 Trending papers & reports
|
> Private LLM inference runs 70B parameter models on consumer Blackwell GPUs using homomorphic encryption, enabling users to query AI without revealing their data to the server.
LINK
|
|
> Transformer attention patterns mathematically favor tokens at the start and end of sequences from initialization, explaining why models struggle with information buried in the middle of long contexts.
LINK
|
|
> A digital fruit fly brain with 139,255 neurons and 54.5 million connections successfully controls a simulated fly body to walk, demonstrating that connectome maps can produce coordinated movement.
LINK
|
|
|
|
🎓 Want to master the AI tools we cover every day?
Techpresso's AI Academy has 300+ step-by-step tutorials on ChatGPT, Claude, Perplexity, and every tool that matters. No fluff — just practical workflows you can use at work. Try it free for 14 days.
|
|
💬 How did you find today's edition?
We read every reply — just reply to this email and let us know how we can improve!
|
|
|
