Blood test predicts Alzheimer’s onset & Generative AI enters music apps - Tech News (Feb 19, 2026)
February 19, 202613m 32s
Audio is streamed directly from the publisher (mcdn.podbean.com) as published in their RSS feed. Play Podcasts does not host this file. Rights-holders can request removal through the copyright & takedown page.
Show Notes
Please support this podcast by checking out our sponsors:
- Consensus: AI for Research. Get a free month - https://get.consensus.app/automated_daily
- Build Any Form, Without Code with Fillout. 50% extra signup credits - https://try.fillout.com/the_automated_daily
- KrispCall: Agentic Cloud Telephony - https://try.krispcall.com/tad
Support The Automated Daily directly:
Buy me a coffee: https://buymeacoffee.com/theautomateddaily
Today's topics: Blood test predicts Alzheimer’s onset - Washington University researchers published a Nature Medicine study using plasma p‑tau217 biomarkers to predict Alzheimer’s symptom onset within ~3–4 years, enabling faster clinical trials and earlier intervention planning. Generative AI enters music apps - Google Gemini is rolling out Lyria 3 for 30‑second music generation plus AI cover art, while Apple Music adds Playlist Playground in iOS 26.4—raising fresh copyright and licensing questions for generative music. Meta’s AI chip build-out - Meta expanded a multiyear Nvidia deal to deploy millions of AI chips, including standalone Grace CPUs and next-gen Vera Rubin racks—part of a massive 2026 capex push for data centers and frontier models. Agentic coding hits GitHub Actions - GitHub Agentic Workflows, in technical preview, runs coding agents inside GitHub Actions with guardrails like read-only defaults, safe outputs, and auditing—aimed at issue triage, docs upkeep, CI debugging, and repo health reports. AI safety, law, and platforms - The UK proposes a 48-hour takedown rule for non-consensual intimate images with major fines, while Meta faces a landmark youth-safety trial and the Pentagon clashes with Anthropic over surveillance and autonomous-weapons limits. Payments sovereignty: UK and EU - The ECB argues a digital euro is needed for monetary sovereignty and cheaper merchant payments, as UK banks explore an account-to-account alternative to Visa and Mastercard to improve resilience and reduce fees. Markets rethink moats in AI - New essays argue ‘rocketship’ career picking is unreliable, software moats are shifting toward scarce industry positions, and AI forecasting splits into empiricists vs extrapolators—while SaaS stocks react to ‘AI eating software’ fears. Defense tech: longer-range missiles - Analysts say Russia is increasingly fielding long-range R‑37M air-to-air missiles on Su‑35 fighters, expanding theoretical threats at distance and complicating NATO air-operations planning. FDA reopens Moderna flu review - The FDA reversed a refusal-to-file and will review Moderna’s mRNA flu vaccine application under a revised filing strategy, with an approval decision expected by August 5, 2026—amid political and regulatory scrutiny of mRNA. Open-source agents and security - A hands-on OpenClaw experiment highlights how ‘skills’ could replace many apps, but also flags observability gaps and a serious exposure issue: tens of thousands of open Gateway instances reportedly leaked keys and widened attack surfaces.
Episode Transcript
Blood test predicts Alzheimer’s onset
We’ll start in health tech, because today’s most concrete “future just arrived” headline is out of Washington University School of Medicine in St. Louis. In a Nature Medicine study published today, researchers describe clock-style models that use a single blood biomarker—plasma p‑tau217—to estimate when a person is likely to develop Alzheimer’s symptoms, typically within about three to four years of accuracy.
The key point: they’re leaning on a known pattern in Alzheimer’s progression—amyloid and tau build up in the brain over time—and showing that blood measurements can act as a practical proxy. The team tested the approach across two cohorts totaling 603 older adults, and importantly, across different testing methods, including at least one FDA-cleared assay in the ADNI dataset. They’ve also released code and a web app for researchers.
This is not a consumer diagnosis tool yet, and it’s not a guarantee for any individual. But for clinical trials—especially prevention trials where timing is everything—being able to identify who’s likely to convert to symptoms in a defined window could dramatically cut costs and speed recruitment compared to PET scans or spinal taps.
Generative AI enters music apps
Staying in biomed policy for a moment: the FDA has reversed its recent refusal to review Moderna’s mRNA flu vaccine application. After a formal Type A meeting, the agency agreed to accept a revised regulatory approach and move the filing forward.
The dispute appears to have centered less on Moderna’s vaccine itself, and more on the Phase 3 trial’s comparator choice—standard-dose flu shots, rather than what regulators argued is the best-available standard of care in older groups. Moderna’s compromise is essentially to split the filing: pursue traditional approval for ages 50 to 64, and accelerated approval for 65-plus, paired with a confirmatory effectiveness trial.
Assuming the review proceeds on schedule, the FDA decision is expected by August 5th, 2026. The broader subtext here is that mRNA has become politically charged again, and that has real consequences for investment and planning across vaccine R&D.
Meta’s AI chip build-out
Now to consumer AI, where the next battleground seems to be… music.
Google says Gemini can generate 30-second music tracks from text prompts, photos, or even user-uploaded video, using a new DeepMind model called Lyria 3. Users can ask for instrumentals or add lyrics, and Google is also pairing it with an image model—yes, really named “Nano Banana”—to create cover art for sharing. Rollout starts on Gemini desktop, then hits mobile shortly after, with availability for users 18 and older in multiple languages.
Google is putting clear limits on free usage—about 10 track generations a day—while paid tiers get more. And notably, Google says users have rights to use the tracks they generate.
Apple, meanwhile, is taking a different angle: it’s not generating songs, it’s generating playlists. A new Apple Music feature called Playlist Playground, bundled into iOS 26.4, turns prompts into a curated playlist with cover art, a description, and 25 songs. That’s a direct shot at similar prompt-based playlist features, including the one Spotify has been experimenting with.
The immediate market reaction was telling—Spotify shares briefly gave up gains after Google’s music-gen news—but analysts don’t see it as existential. The real story is that AI creation tools are sliding into mainstream consumer surfaces, which forces the music industry to keep wrestling with licensing, training data, and what “original” even means. Google says it’s using filters, blocks obvious “make it like artist X” lifting, and claims Lyria 3 is trained on music it has rights to use under YouTube and partner agreements. Expect that claim to be tested—legally and culturally.
Agentic coding hits GitHub Actions
On AI hardware and scale: Meta expanded a multiyear partnership with Nvidia to deploy what it calls “millions” of AI chips across its data-center build-out. The headline details include next-gen GPUs, rack-scale Vera Rubin systems, and—importantly—Nvidia’s Grace CPUs used at large scale as standalone data-center chips.
No price tag was announced, but analysts are reading this as a tens-of-billions type commitment, lining up with Meta’s plan to spend up to $135 billion on AI in 2026. Meta also highlighted networking gear like Spectrum-X Ethernet, and even Nvidia security capabilities tied to AI features on WhatsApp.
The practical takeaway: demand is still running ahead of supply for top-tier AI compute, and big platforms are locking in allocation. Meta says it’s not exclusively dependent on Nvidia—it’s designing its own silicon and using AMD too—but this is a very clear “we’re buying capacity and co-designing for it” signal.
And in a smaller-but-related Meta note: reports say the company is again exploring a smartwatch for 2026, code-named Malibu 2, potentially with a built-in Meta AI assistant. Meta already has smart glasses momentum; a watch could become the wrist controller for that ecosystem—or it could be a very expensive experiment. The article’s phrasing was apt: execution matters, or it risks a Fire Phone-style flop.
AI safety, law, and platforms
Let’s shift to AI agents and software development, where the tools are getting more “hands-on” inside existing workflows.
GitHub has introduced Agentic Workflows in technical preview—coding agents running inside GitHub Actions. The pitch is “Continuous AI”: you describe outcomes in plain Markdown, and an agent does the repo chores that don’t fit neatly into deterministic YAML. Think issue triage, docs maintenance, simplifying code, improving tests, investigating CI failures, and producing periodic health reports.
What’s notable is the security posture. GitHub is leaning hard on defense-in-depth: read-only by default, explicit approvals for write actions, safe outputs like pull requests or issue comments, tool allowlists, and auditing. Also, they’re not locking you into one model—workflows can be configured to use engines like Copilot CLI, Claude Code, or OpenAI Codex.
At the same time, a cautionary counterpoint came from developer and data tooling leader Wes McKinney, who’s been writing about “agentic” coding changing his habits—sometimes literally his sleep schedule. His point isn’t “agents are bad,” it’s that cheap code generation can create an ‘agentic tar pit’: more code, more scope creep, more architectural drift, and a bigger human burden to preserve conceptual integrity. In other words, the bottleneck moves upward—from typing code to making good decisions about what the system should be.
That theme also shows up in the observability world. New Relic is running a big virtual event next week on “Intelligent Observability” and agentic operations—very much the “AI should remediate incidents” narrative. Put all of this together and you can see where 2026 is heading: agents everywhere, and an arms race around guardrails, accountability, and cost controls.
Payments sovereignty: UK and EU
On software delivery economics, one concrete case study making the rounds: Drata says it partnered with QA Wolf to accelerate regression testing. The headline metrics are dramatic—expanding automated coverage from roughly 100 to over 400 test cases, while shrinking QA feedback time to around 10 to 15 minutes through parallel execution, and claiming more than $500,000 a year saved.
Whether those exact numbers generalize will vary by org, but the direction is real: QA is becoming a prime target for “agent-plus-service” models—because flakey tests and constant maintenance are exactly the kind of grind that teams want to offload.
Markets rethink moats in AI
Now, regulation and safety—three stories that all rhyme.
First, the UK government is proposing a new law to require platforms to remove intimate images shared without consent within 48 hours, and to prevent re-uploads. The proposed penalties are substantial—up to 10% of global sales, or even service blocking in the UK. The political message is straightforward: treat intimate image abuse with the seriousness of terrorist content and child sexual abuse material, and stop forcing victims to play platform-by-platform “whack-a-mole.”
Second, in the U.S., Meta CEO Mark Zuckerberg testified in a major social media safety trial that some observers are calling a “Big Tobacco” moment for the industry. The case centers on claims that product design choices contributed to addiction-like usage and mental health harm. One detail that stood out: court attention on cosmetic-style filters and internal debate over whether restricting them was protective or overly paternalistic. Whatever the verdict, the trial is building a public record—emails, targets, and decision rationales—that will shape future regulation.
Third, AI policy is getting more openly political. Axios reports a dispute between the U.S. Department of Defense and Anthropic during talks over letting the Pentagon use Anthropic tech on classified systems. The flashpoint is reportedly Anthropic’s insistence on limits—no domestic mass surveillance of Americans, and no autonomous weapons without a human in the loop. Pentagon leadership, per the report, viewed that as the company trying to dictate military use.
These three stories all point to the same tension: the most capable platforms and models are becoming infrastructure, but society is still arguing about where the red lines should be—and who gets to draw them.
Defense tech: longer-range missiles
Let’s talk money and strategy, because the market narrative around AI is shifting again.
One widely discussed analysis argues that “picking rocketships” is becoming a bad career strategy in tech. The reason is simple: with fast capital, crowded markets, and AI turning features into commodities, it’s harder to tell which company will be the winner. The proposed litmus test is refreshingly practical: if the company struggles or fails, will you still be glad you took the job—for the skills and relationships you built?
A separate “moats in the age of AI” argument says we’re obsessing over whether software is defensible, when the real value increasingly concentrates around scarce positions—control of complementary assets, distribution, regulated footholds, and industry choke points. In that view, abundant tech inputs like AI don’t flatten outcomes; they can widen the gap between leaders and everyone else.
And that frames today’s public-market anxiety: software equities have seen a brutal drawdown, with investors spooked by rapid AI product rollouts from companies like Anthropic and OpenAI—and the broader ‘AI eating SaaS’ story. But the more sober conclusion is: SaaS isn’t dead; it’s being repriced and reorganized. The fragile part is hyper-narrow tools that can be replicated quickly, or incumbents that move slowly to match AI capabilities—especially if agent layers start displacing the ‘front door’ of enterprise software.
FDA reopens Moderna flu review
Payments and sovereignty are another place where “infrastructure politics” is suddenly mainstream.
At the European Central Bank, executive board member Piero Cipollone argued today that a digital euro is needed to keep central bank money usable as payments go digital and cash use declines. The ECB’s line is that Europe is overly dependent on non-European payment providers, and a digital euro could reduce costs for merchants while preserving inclusion and offline privacy.
In the UK, bank executives are reportedly meeting under an industry-backed effort—overseen by the Bank of England—to explore a national payment alternative to Visa and Mastercard. The concept is account-to-account payments, moving money directly from bank accounts to merchants without card rails. Supporters say it could improve resilience and lower costs; critics will point out that building widely adopted payment networks is hard, and incentives have to line up for banks, merchants, and consumers.
Either way, you can hear the same theme in London and Frankfurt: payments aren’t just fintech. They’re strategic infrastructure.
Open-source agents and security
Two quick final notes.
On defense tech: a RUSI analyst says Russia is increasingly arming Su-35 fighters with the long-range R‑37M air-to-air missile. The estimated reach is vastly beyond older loadouts—though real-world effectiveness depends on conditions and countermeasures. The strategic point is that longer-range weapons, paired with integrated air defenses, complicate planning for air operations.
And on open-source agents: a Substack deep dive into OpenClaw—a minimalist agent system—argues that “skills” could replace many everyday apps by turning capabilities into modular contracts. But it also flags a serious security problem: a report of 30,000-plus exposed OpenClaw Gateway instances in a short window, with readable keys and open ports. It’s a reminder that ‘agentic’ convenience can create new blast radiuses when defaults and deployment hygiene don’t keep up.
Subscribe to edition specific feeds:
- Space news
* Apple Podcast English
* Spotify English
* RSS English Spanish French
- Top news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French
- Tech news
* Apple Podcast English Spanish French
* Spotify English Spanish Spanish
* RSS English Spanish French
- Hacker news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French
- AI news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French
Visit our website at https://theautomateddaily.com/
Send feedback to [email protected]
Youtube
LinkedIn
X (Twitter)
- Consensus: AI for Research. Get a free month - https://get.consensus.app/automated_daily
- Build Any Form, Without Code with Fillout. 50% extra signup credits - https://try.fillout.com/the_automated_daily
- KrispCall: Agentic Cloud Telephony - https://try.krispcall.com/tad
Support The Automated Daily directly:
Buy me a coffee: https://buymeacoffee.com/theautomateddaily
Today's topics: Blood test predicts Alzheimer’s onset - Washington University researchers published a Nature Medicine study using plasma p‑tau217 biomarkers to predict Alzheimer’s symptom onset within ~3–4 years, enabling faster clinical trials and earlier intervention planning. Generative AI enters music apps - Google Gemini is rolling out Lyria 3 for 30‑second music generation plus AI cover art, while Apple Music adds Playlist Playground in iOS 26.4—raising fresh copyright and licensing questions for generative music. Meta’s AI chip build-out - Meta expanded a multiyear Nvidia deal to deploy millions of AI chips, including standalone Grace CPUs and next-gen Vera Rubin racks—part of a massive 2026 capex push for data centers and frontier models. Agentic coding hits GitHub Actions - GitHub Agentic Workflows, in technical preview, runs coding agents inside GitHub Actions with guardrails like read-only defaults, safe outputs, and auditing—aimed at issue triage, docs upkeep, CI debugging, and repo health reports. AI safety, law, and platforms - The UK proposes a 48-hour takedown rule for non-consensual intimate images with major fines, while Meta faces a landmark youth-safety trial and the Pentagon clashes with Anthropic over surveillance and autonomous-weapons limits. Payments sovereignty: UK and EU - The ECB argues a digital euro is needed for monetary sovereignty and cheaper merchant payments, as UK banks explore an account-to-account alternative to Visa and Mastercard to improve resilience and reduce fees. Markets rethink moats in AI - New essays argue ‘rocketship’ career picking is unreliable, software moats are shifting toward scarce industry positions, and AI forecasting splits into empiricists vs extrapolators—while SaaS stocks react to ‘AI eating software’ fears. Defense tech: longer-range missiles - Analysts say Russia is increasingly fielding long-range R‑37M air-to-air missiles on Su‑35 fighters, expanding theoretical threats at distance and complicating NATO air-operations planning. FDA reopens Moderna flu review - The FDA reversed a refusal-to-file and will review Moderna’s mRNA flu vaccine application under a revised filing strategy, with an approval decision expected by August 5, 2026—amid political and regulatory scrutiny of mRNA. Open-source agents and security - A hands-on OpenClaw experiment highlights how ‘skills’ could replace many apps, but also flags observability gaps and a serious exposure issue: tens of thousands of open Gateway instances reportedly leaked keys and widened attack surfaces.
Episode Transcript
Blood test predicts Alzheimer’s onset
We’ll start in health tech, because today’s most concrete “future just arrived” headline is out of Washington University School of Medicine in St. Louis. In a Nature Medicine study published today, researchers describe clock-style models that use a single blood biomarker—plasma p‑tau217—to estimate when a person is likely to develop Alzheimer’s symptoms, typically within about three to four years of accuracy.
The key point: they’re leaning on a known pattern in Alzheimer’s progression—amyloid and tau build up in the brain over time—and showing that blood measurements can act as a practical proxy. The team tested the approach across two cohorts totaling 603 older adults, and importantly, across different testing methods, including at least one FDA-cleared assay in the ADNI dataset. They’ve also released code and a web app for researchers.
This is not a consumer diagnosis tool yet, and it’s not a guarantee for any individual. But for clinical trials—especially prevention trials where timing is everything—being able to identify who’s likely to convert to symptoms in a defined window could dramatically cut costs and speed recruitment compared to PET scans or spinal taps.
Generative AI enters music apps
Staying in biomed policy for a moment: the FDA has reversed its recent refusal to review Moderna’s mRNA flu vaccine application. After a formal Type A meeting, the agency agreed to accept a revised regulatory approach and move the filing forward.
The dispute appears to have centered less on Moderna’s vaccine itself, and more on the Phase 3 trial’s comparator choice—standard-dose flu shots, rather than what regulators argued is the best-available standard of care in older groups. Moderna’s compromise is essentially to split the filing: pursue traditional approval for ages 50 to 64, and accelerated approval for 65-plus, paired with a confirmatory effectiveness trial.
Assuming the review proceeds on schedule, the FDA decision is expected by August 5th, 2026. The broader subtext here is that mRNA has become politically charged again, and that has real consequences for investment and planning across vaccine R&D.
Meta’s AI chip build-out
Now to consumer AI, where the next battleground seems to be… music.
Google says Gemini can generate 30-second music tracks from text prompts, photos, or even user-uploaded video, using a new DeepMind model called Lyria 3. Users can ask for instrumentals or add lyrics, and Google is also pairing it with an image model—yes, really named “Nano Banana”—to create cover art for sharing. Rollout starts on Gemini desktop, then hits mobile shortly after, with availability for users 18 and older in multiple languages.
Google is putting clear limits on free usage—about 10 track generations a day—while paid tiers get more. And notably, Google says users have rights to use the tracks they generate.
Apple, meanwhile, is taking a different angle: it’s not generating songs, it’s generating playlists. A new Apple Music feature called Playlist Playground, bundled into iOS 26.4, turns prompts into a curated playlist with cover art, a description, and 25 songs. That’s a direct shot at similar prompt-based playlist features, including the one Spotify has been experimenting with.
The immediate market reaction was telling—Spotify shares briefly gave up gains after Google’s music-gen news—but analysts don’t see it as existential. The real story is that AI creation tools are sliding into mainstream consumer surfaces, which forces the music industry to keep wrestling with licensing, training data, and what “original” even means. Google says it’s using filters, blocks obvious “make it like artist X” lifting, and claims Lyria 3 is trained on music it has rights to use under YouTube and partner agreements. Expect that claim to be tested—legally and culturally.
Agentic coding hits GitHub Actions
On AI hardware and scale: Meta expanded a multiyear partnership with Nvidia to deploy what it calls “millions” of AI chips across its data-center build-out. The headline details include next-gen GPUs, rack-scale Vera Rubin systems, and—importantly—Nvidia’s Grace CPUs used at large scale as standalone data-center chips.
No price tag was announced, but analysts are reading this as a tens-of-billions type commitment, lining up with Meta’s plan to spend up to $135 billion on AI in 2026. Meta also highlighted networking gear like Spectrum-X Ethernet, and even Nvidia security capabilities tied to AI features on WhatsApp.
The practical takeaway: demand is still running ahead of supply for top-tier AI compute, and big platforms are locking in allocation. Meta says it’s not exclusively dependent on Nvidia—it’s designing its own silicon and using AMD too—but this is a very clear “we’re buying capacity and co-designing for it” signal.
And in a smaller-but-related Meta note: reports say the company is again exploring a smartwatch for 2026, code-named Malibu 2, potentially with a built-in Meta AI assistant. Meta already has smart glasses momentum; a watch could become the wrist controller for that ecosystem—or it could be a very expensive experiment. The article’s phrasing was apt: execution matters, or it risks a Fire Phone-style flop.
AI safety, law, and platforms
Let’s shift to AI agents and software development, where the tools are getting more “hands-on” inside existing workflows.
GitHub has introduced Agentic Workflows in technical preview—coding agents running inside GitHub Actions. The pitch is “Continuous AI”: you describe outcomes in plain Markdown, and an agent does the repo chores that don’t fit neatly into deterministic YAML. Think issue triage, docs maintenance, simplifying code, improving tests, investigating CI failures, and producing periodic health reports.
What’s notable is the security posture. GitHub is leaning hard on defense-in-depth: read-only by default, explicit approvals for write actions, safe outputs like pull requests or issue comments, tool allowlists, and auditing. Also, they’re not locking you into one model—workflows can be configured to use engines like Copilot CLI, Claude Code, or OpenAI Codex.
At the same time, a cautionary counterpoint came from developer and data tooling leader Wes McKinney, who’s been writing about “agentic” coding changing his habits—sometimes literally his sleep schedule. His point isn’t “agents are bad,” it’s that cheap code generation can create an ‘agentic tar pit’: more code, more scope creep, more architectural drift, and a bigger human burden to preserve conceptual integrity. In other words, the bottleneck moves upward—from typing code to making good decisions about what the system should be.
That theme also shows up in the observability world. New Relic is running a big virtual event next week on “Intelligent Observability” and agentic operations—very much the “AI should remediate incidents” narrative. Put all of this together and you can see where 2026 is heading: agents everywhere, and an arms race around guardrails, accountability, and cost controls.
Payments sovereignty: UK and EU
On software delivery economics, one concrete case study making the rounds: Drata says it partnered with QA Wolf to accelerate regression testing. The headline metrics are dramatic—expanding automated coverage from roughly 100 to over 400 test cases, while shrinking QA feedback time to around 10 to 15 minutes through parallel execution, and claiming more than $500,000 a year saved.
Whether those exact numbers generalize will vary by org, but the direction is real: QA is becoming a prime target for “agent-plus-service” models—because flakey tests and constant maintenance are exactly the kind of grind that teams want to offload.
Markets rethink moats in AI
Now, regulation and safety—three stories that all rhyme.
First, the UK government is proposing a new law to require platforms to remove intimate images shared without consent within 48 hours, and to prevent re-uploads. The proposed penalties are substantial—up to 10% of global sales, or even service blocking in the UK. The political message is straightforward: treat intimate image abuse with the seriousness of terrorist content and child sexual abuse material, and stop forcing victims to play platform-by-platform “whack-a-mole.”
Second, in the U.S., Meta CEO Mark Zuckerberg testified in a major social media safety trial that some observers are calling a “Big Tobacco” moment for the industry. The case centers on claims that product design choices contributed to addiction-like usage and mental health harm. One detail that stood out: court attention on cosmetic-style filters and internal debate over whether restricting them was protective or overly paternalistic. Whatever the verdict, the trial is building a public record—emails, targets, and decision rationales—that will shape future regulation.
Third, AI policy is getting more openly political. Axios reports a dispute between the U.S. Department of Defense and Anthropic during talks over letting the Pentagon use Anthropic tech on classified systems. The flashpoint is reportedly Anthropic’s insistence on limits—no domestic mass surveillance of Americans, and no autonomous weapons without a human in the loop. Pentagon leadership, per the report, viewed that as the company trying to dictate military use.
These three stories all point to the same tension: the most capable platforms and models are becoming infrastructure, but society is still arguing about where the red lines should be—and who gets to draw them.
Defense tech: longer-range missiles
Let’s talk money and strategy, because the market narrative around AI is shifting again.
One widely discussed analysis argues that “picking rocketships” is becoming a bad career strategy in tech. The reason is simple: with fast capital, crowded markets, and AI turning features into commodities, it’s harder to tell which company will be the winner. The proposed litmus test is refreshingly practical: if the company struggles or fails, will you still be glad you took the job—for the skills and relationships you built?
A separate “moats in the age of AI” argument says we’re obsessing over whether software is defensible, when the real value increasingly concentrates around scarce positions—control of complementary assets, distribution, regulated footholds, and industry choke points. In that view, abundant tech inputs like AI don’t flatten outcomes; they can widen the gap between leaders and everyone else.
And that frames today’s public-market anxiety: software equities have seen a brutal drawdown, with investors spooked by rapid AI product rollouts from companies like Anthropic and OpenAI—and the broader ‘AI eating SaaS’ story. But the more sober conclusion is: SaaS isn’t dead; it’s being repriced and reorganized. The fragile part is hyper-narrow tools that can be replicated quickly, or incumbents that move slowly to match AI capabilities—especially if agent layers start displacing the ‘front door’ of enterprise software.
FDA reopens Moderna flu review
Payments and sovereignty are another place where “infrastructure politics” is suddenly mainstream.
At the European Central Bank, executive board member Piero Cipollone argued today that a digital euro is needed to keep central bank money usable as payments go digital and cash use declines. The ECB’s line is that Europe is overly dependent on non-European payment providers, and a digital euro could reduce costs for merchants while preserving inclusion and offline privacy.
In the UK, bank executives are reportedly meeting under an industry-backed effort—overseen by the Bank of England—to explore a national payment alternative to Visa and Mastercard. The concept is account-to-account payments, moving money directly from bank accounts to merchants without card rails. Supporters say it could improve resilience and lower costs; critics will point out that building widely adopted payment networks is hard, and incentives have to line up for banks, merchants, and consumers.
Either way, you can hear the same theme in London and Frankfurt: payments aren’t just fintech. They’re strategic infrastructure.
Open-source agents and security
Two quick final notes.
On defense tech: a RUSI analyst says Russia is increasingly arming Su-35 fighters with the long-range R‑37M air-to-air missile. The estimated reach is vastly beyond older loadouts—though real-world effectiveness depends on conditions and countermeasures. The strategic point is that longer-range weapons, paired with integrated air defenses, complicate planning for air operations.
And on open-source agents: a Substack deep dive into OpenClaw—a minimalist agent system—argues that “skills” could replace many everyday apps by turning capabilities into modular contracts. But it also flags a serious security problem: a report of 30,000-plus exposed OpenClaw Gateway instances in a short window, with readable keys and open ports. It’s a reminder that ‘agentic’ convenience can create new blast radiuses when defaults and deployment hygiene don’t keep up.
Subscribe to edition specific feeds:
- Space news
* Apple Podcast English
* Spotify English
* RSS English Spanish French
- Top news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French
- Tech news
* Apple Podcast English Spanish French
* Spotify English Spanish Spanish
* RSS English Spanish French
- Hacker news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French
- AI news
* Apple Podcast English Spanish French
* Spotify English Spanish French
* RSS English Spanish French
Visit our website at https://theautomateddaily.com/
Send feedback to [email protected]
Youtube
X (Twitter)