Riley on the day · June 26th

Most people perceive a cork in a bottle of wine as a simple plug meant to keep the liquid in and the outside world out. In the recent study published in Science Advances, a team of French scientists demonstrated the cork is way more than that. By regulating the oxygen transfer into and out of the wine bottle, it works almost as another ingredient. “Twenty years ago, our group focused on the oxidation and aging of wine and all its parameters,” Thomas Karbowiak said. “Oxygen diffusion through cork stoppers is one of these parameters.” Karbowiak is a chemist at the University of Burgundy, France, and the senior author of the study. The mini-bottle experiment Oxidation is one of the key drivers of wine aging. A slow, limited ingress of oxygen helps wine mature, smoothing out harsh tannins and bringing out an aromatic complexity. But when too much oxygen gets into the bottle too quickly, it can make the wine stale, brownish in color, and unpleasant to drink. That’s because it will also react with alcohol and phenols in the same process that makes a cut apple turn brown. Read full article

The French team set up tiny bottles with real corks and watched oxygen drift in and out, treating the cork not as a passive seal but as an active participant in the wine’s chemistry. By measuring how the cork’s porous structure lets a trickle of air through, they showed that the stopper subtly controls the rate at which the wine ages, almost like a hidden ingredient that can be tuned.

What they found is that a slow, steady seep of oxygen smooths tannins and deepens aroma, while a sudden burst pushes the wine toward browning and stale flavors. The cork’s tiny channels act like a valve, letting just enough air in to keep the wine evolving without overwhelming it. The experiment also highlighted how different cork qualities change that balance, suggesting that the choice of stopper could be as deliberate as the grape blend.

In practical terms, the study reminds winemakers that the cork isn’t just a barrier; it’s a dynamic filter that shapes the wine’s journey from bottle to glass. By understanding the mechanics of oxygen diffusion, producers can better predict how a wine will develop over time and perhaps select corks that match the desired aging profile. It’s a reminder that even the smallest piece of the system can have a surprisingly big influence.

The French team ran a controlled lab experiment—tiny wine bottles sealed with real corks—so they could watch oxygen move in and out. By measuring the gas flow over weeks, they showed the cork isn’t just a plug; it acts like a slow‑release valve, letting just enough oxygen seep through to guide the wine’s aging.

That modest, steady drift of oxygen smooths tannins and builds aroma, which is why many wines develop that pleasant complexity over time. When the cork lets too much oxygen in too quickly, the wine turns brownish and flat, much like a cut apple that browns fast.

The takeaway is simple: the cork’s permeability is a key piece of the aging puzzle, and tweaking that property could help winemakers fine‑tune how a wine matures, without adding anything else to the bottle.

Claude Tag is Anthropic’s latest attempt at getting Claude out of your DMs and into your team’s Slack channels.

AI assistants are increasingly showing up in the workplace to perform research, coding, writing, and analysis, but the results of those interactions typically remains tied to individual conversations rather than being shared across projects and teams.

That limitation is what Anthropic is addressing with Claude Tag , a new Slack channel-based experience for its Enterprise and Team customers, designed to give them a shared AI collaborator that retains context across conversations and participates in work with multiple employees.

Tag will replace Anthropic’s previous attempt at this, Claude in Slack, would only interact with one person (although it’s responses were visible to all in a channel) and its context was limited to the last 20 messages in a channel.

Claude Tag has a much larger context and can be asked to complete tasks on its own, returning with results and a log of how it completed the task for review. It can also schedule follow-up work for itself, enabling projects to continue over hours or days without constant prompting, Anthropic said.

Tag also has an “ambient” mode: when this is enabled, it proactively surfaces relevant information from other channels and connected tools, notifying teams about updates that may be important, and following up on unresolved discussions or tasks, the company said.

These features could act as an immediate productivity enhancer for enterprises by reducing coordination overhead and improving collaboration across engineering, developer, and business teams, analysts said.

The biggest benefit for enterprises is the reduction in time spent finding information and rebuilding context across AI interactions, according to Pareekh Jain , principal analyst at Pareekh Consulting. “Because Claude remembers what’s been said across channels, it acts like shared team memory, so no one has to repeat context or hold endless catch-up meetings.”

That reduction in coordination overhead, according to Amit Jena , AI development manager at IT consulting firm Kanerika, could deliver productivity gains that go well beyond the incremental improvements associated with traditional AI assistants.

“For engineering teams, Claude Tag will help reduce time spent on debugging through fragmented Slack discussions, summarizing long incident threads, pulling context across repos, tickets, and logs, and documenting decisions after the fact,” Jena said, while for business teams, “It could enable faster decision-making from thread summaries while reducing follow-ups in cross-functional work.”

Sohail Dev Majumdar , principal analyst at Gartner, though, sees greater benefits than mere productivity gains, particularly for CIOs and other technology leaders.

The new offering reflects growing demand among enterprises for AI systems that can work across teams, retain organizational context, participate more actively in day-to-day workflows, and generate more measurable return on investment, he said.

On that last point, though, he warned that CIOs will need to change how they measure ROI for collaborative AI systems compared to traditional AI assistants: “ROI measurement must go beyond license counts, focusing on both hard metrics like time savings and error reduction; and soft metrics, such as employee satisfaction and innovation.”

Jena said CIOs will also need to reconsider auditability and governance around Tag, as it can access context, data, and tools outside individual user boundaries and influence downstream systems.

“CIOs should rethink who can assign tasks to AI agents, what data a channel-level agent can access, how AI-generated outputs are reviewed and approved, how long conversational memory should persist, and. how compliance logs map to AI actions,” Jena said.

With this in mind, Anthropic is shipping controls that will enable system administrators to filter access to data, tools, and Slack channels along with spending limits. These controls, the company said, will also enable administrators to create separate Claude instances for different teams, with each instance limited to the channels and information assigned to it.

To encourage adoption, Anthropic is offering a one-time pool of launch credits to eligible organizations, enabling employees to experiment with the service before it begins consuming their regular usage allocation. Eligible Claude Enterprise customers will receive $25,000 in promotional credits, while qualifying Claude Team customers with at least 10 paid seats will receive credits worth $2,500, the company said.

The credits can be used only for Claude Tag interactions in Slack channels and will remain valid through September 1, 2026.

Tag will replace Claude in Slack on August 3, 2026 — or administrators can opt in early in the next 30 days.

Claude Tag is Anthropic’s latest attempt at getting Claude out of your DMs and into your team’s Slack channels.

AI assistants are increasingly showing up in the workplace to perform research, coding, writing, and analysis, but the results of those interactions typically remains tied to individual conversations rather than being shared across projects and teams.

That limitation is what Anthropic is addressing with Claude Tag , a new Slack channel-based experience for its Enterprise and Team customers, designed to give them a shared AI collaborator that retains context across conversations and participates in work with multiple employees.

Tag will replace Anthropic’s previous attempt at this, Claude in Slack, would only interact with one person (although it’s responses were visible to all in a channel) and its context was limited to the last 20 messages in a channel.

Claude Tag has a much larger context and can be asked to complete tasks on its own, returning with results and a log of how it completed the task for review. It can also schedule follow-up work for itself, enabling projects to continue over hours or days without constant prompting, Anthropic said.

Tag also has an “ambient” mode: when this is enabled, it proactively surfaces relevant information from other channels and connected tools, notifying teams about updates that may be important, and following up on unresolved discussions or tasks, the company said.

These features could act as an immediate productivity enhancer for enterprises by reducing coordination overhead and improving collaboration across engineering, developer, and business teams, analysts said.

The biggest benefit for enterprises is the reduction in time spent finding information and rebuilding context across AI interactions, according to Pareekh Jain , principal analyst at Pareekh Consulting. “Because Claude remembers what’s been said across channels, it acts like shared team memory, so no one has to repeat context or hold endless catch-up meetings.”

That reduction in coordination overhead, according to Amit Jena , AI development manager at IT consulting firm Kanerika, could deliver productivity gains that go well beyond the incremental improvements associated with traditional AI assistants.

“For engineering teams, Claude Tag will help reduce time spent on debugging through fragmented Slack discussions, summarizing long incident threads, pulling context across repos, tickets, and logs, and documenting decisions after the fact,” Jena said, while for business teams, “It could enable faster decision-making from thread summaries while reducing follow-ups in cross-functional work.”

Sohail Dev Majumdar , principal analyst at Gartner, though, sees greater benefits than mere productivity gains, particularly for CIOs and other technology leaders.

The new offering reflects growing demand among enterprises for AI systems that can work across teams, retain organizational context, participate more actively in day-to-day workflows, and generate more measurable return on investment, he said.

On that last point, though, he warned that CIOs will need to change how they measure ROI for collaborative AI systems compared to traditional AI assistants: “ROI measurement must go beyond license counts, focusing on both hard metrics like time savings and error reduction; and soft metrics, such as employee satisfaction and innovation.”

Jena said CIOs will also need to reconsider auditability and governance around Tag, as it can access context, data, and tools outside individual user boundaries and influence downstream systems.

“CIOs should rethink who can assign tasks to AI agents, what data a channel-level agent can access, how AI-generated outputs are reviewed and approved, how long conversational memory should persist, and. how compliance logs map to AI actions,” Jena said.

With this in mind, Anthropic is shipping controls that will enable system administrators to filter access to data, tools, and Slack channels along with spending limits. These controls, the company said, will also enable administrators to create separate Claude instances for different teams, with each instance limited to the channels and information assigned to it.

To encourage adoption, Anthropic is offering a one-time pool of launch credits to eligible organizations, enabling employees to experiment with the service before it begins consuming their regular usage allocation. Eligible Claude Enterprise customers will receive $25,000 in promotional credits, while qualifying Claude Team customers with at least 10 paid seats will receive credits worth $2,500, the company said.

The credits can be used only for Claude Tag interactions in Slack channels and will remain valid through September 1, 2026.

Tag will replace Claude in Slack on August 3, 2026 — or administrators can opt in early in the next 30 days.

Claude Tag is Anthropic’s latest attempt at getting Claude out of your DMs and into your team’s Slack channels.

AI assistants are increasingly showing up in the workplace to perform research, coding, writing, and analysis, but the results of those interactions typically remains tied to individual conversations rather than being shared across projects and teams.

That limitation is what Anthropic is addressing with Claude Tag , a new Slack channel-based experience for its Enterprise and Team customers, designed to give them a shared AI collaborator that retains context across conversations and participates in work with multiple employees.

Tag will replace Anthropic’s previous attempt at this, Claude in Slack, would only interact with one person (although it’s responses were visible to all in a channel) and its context was limited to the last 20 messages in a channel.

Claude Tag has a much larger context and can be asked to complete tasks on its own, returning with results and a log of how it completed the task for review. It can also schedule follow-up work for itself, enabling projects to continue over hours or days without constant prompting, Anthropic said.

Tag also has an “ambient” mode: when this is enabled, it proactively surfaces relevant information from other channels and connected tools, notifying teams about updates that may be important, and following up on unresolved discussions or tasks, the company said.

These features could act as an immediate productivity enhancer for enterprises by reducing coordination overhead and improving collaboration across engineering, developer, and business teams, analysts said.

The biggest benefit for enterprises is the reduction in time spent finding information and rebuilding context across AI interactions, according to Pareekh Jain , principal analyst at Pareekh Consulting. “Because Claude remembers what’s been said across channels, it acts like shared team memory, so no one has to repeat context or hold endless catch-up meetings.”

That reduction in coordination overhead, according to Amit Jena , AI development manager at IT consulting firm Kanerika, could deliver productivity gains that go well beyond the incremental improvements associated with traditional AI assistants.

“For engineering teams, Claude Tag will help reduce time spent on debugging through fragmented Slack discussions, summarizing long incident threads, pulling context across repos, tickets, and logs, and documenting decisions after the fact,” Jena said, while for business teams, “It could enable faster decision-making from thread summaries while reducing follow-ups in cross-functional work.”

Sohail Dev Majumdar , principal analyst at Gartner, though, sees greater benefits than mere productivity gains, particularly for CIOs and other technology leaders.

The new offering reflects growing demand among enterprises for AI systems that can work across teams, retain organizational context, participate more actively in day-to-day workflows, and generate more measurable return on investment, he said.

On that last point, though, he warned that CIOs will need to change how they measure ROI for collaborative AI systems compared to traditional AI assistants: “ROI measurement must go beyond license counts, focusing on both hard metrics like time savings and error reduction; and soft metrics, such as employee satisfaction and innovation.”

Jena said CIOs will also need to reconsider auditability and governance around Tag, as it can access context, data, and tools outside individual user boundaries and influence downstream systems.

“CIOs should rethink who can assign tasks to AI agents, what data a channel-level agent can access, how AI-generated outputs are reviewed and approved, how long conversational memory should persist, and. how compliance logs map to AI actions,” Jena said.

With this in mind, Anthropic is shipping controls that will enable system administrators to filter access to data, tools, and Slack channels along with spending limits. These controls, the company said, will also enable administrators to create separate Claude instances for different teams, with each instance limited to the channels and information assigned to it.

To encourage adoption, Anthropic is offering a one-time pool of launch credits to eligible organizations, enabling employees to experiment with the service before it begins consuming their regular usage allocation. Eligible Claude Enterprise customers will receive $25,000 in promotional credits, while qualifying Claude Team customers with at least 10 paid seats will receive credits worth $2,500, the company said.

The credits can be used only for Claude Tag interactions in Slack channels and will remain valid through September 1, 2026.

Tag will replace Claude in Slack on August 3, 2026 — or administrators can opt in early in the next 30 days.

From storyflo. This is your daily audio brief for June 25th.

Brock here. June 25th. Five stories from across the leagues — straight ahead.

Let's get into it.

First, from PFT alt. New York Times calls Dianna Russini enlisting a coach to get out of a ticket "unacceptable conduct".

Dianna Russini told a recent podcast that a police officer pulled her over for texting while her two sons rode in the back seat. She mentioned that Bills coach Sean McDermott had just been fired and that she was trying to break the news.

From storyflo. This is your daily audio brief for June 25th.

It's Riley, June 25th. Ten stories from overnight — sourcing first, takes second.

Let's get into it.

First, from Raw America. Trump Panics as GOP Senators Threaten to Tank His Agenda.

Good morning. I’m Thom Hartmann. Senate Republicans are siding with Majority Leader John Thune over Trump as the president demands senators blow up the rules to pass his agenda.

From storyflo. This is your daily audio brief for June 25th.

It's Mason. June 25th. Five things on the tape worth your attention this morning.

Let's get into it.

First, from drstorm. Why Claude Changed The AI Risk Equation #168b.

We are honored to count you among the >600.000 readers of “DIGITAL STORM weekly”. Please help grow our community by inviting your friends. This week the cost and control of frontier capability stopped being a technical question and became a governance one, on three fronts at once.

From storyflo. This is your daily audio brief for June 25th.

Iris here. June 25th. Health and longevity in ten — let's begin with what changed my mind this week.

Let's get into it.

First, from Dr. Mary Claire. What actually protects your brain in midlife.

The most solid proof comes from two randomized trials. The Finnish FINGER study (2015) and the U.S. POINTER trial (2025) tested a bundle of habits—exercise, a MIND‑style diet, cognitive‑social activity, and vascular monitoring—and both showed modest but reliable gains in memory and thinking scores.