A new project to bring back Vine’s six-second looping videos is now available for download on the App Store and Google Play. Divine, as this Vine reboot is called, offers access to an archive of roughly 500,000 Vine videos, restored from a backup of the original service, and allows creators to post new Vines once again.

Divine was financed by “and Other Stuff,” a nonprofit formed in May 2025 by Twitter co-founder Jack Dorsey. The non-profit is focused on funding experimental open source projects that have the potential to transform the social media landscape. Dorsey’s backing of Divine doesn’t make him a traditional investor since he’s not looking to get a return here. Rather, his goal is to correct an earlier mistake he made as CEO of Twitter: shutting down Vine in the first place.

To create Divine, Evan Henshaw-Plath, an early Twitter employee and member of “and Other Stuff,” explored the Vine archive. Henshaw-Plath, who goes by “Rabble” online, explained that much of Vine’s content was originally backed up by a community archiving project known as the Archive Team.

Image Credits:Divine

Those videos had been stored as large, 40-50 GB binary files, which required Rabble to write big data scripts to figure out how the files worked and how to reconstruct them, along with the user engagement, like the views, likes, and comments, that were associated with the original videos.

Not all data was able to be restored, but progress has been made. The app initially launched to testers last November with some 100,000 of Vine’s top videos, then grew to around 300,000 videos just ahead of today’s launch, Rabble told TechCrunch. Now, the app hosts roughly 500,000 videos from nearly 100,000 original Vine creators as it becomes publicly available for the first time.

The effort has attracted the attention of several early Vine creators, including Lele Pons, JimmyHere, MightyDuck, and Jack and Jack, among others. (Divine user profiles are viewable on the web, even if you don’t have the Divine mobile app.)

Image Credits:Divine

Rabble said the initial plan was to quickly push out the app after some initial tests, but early Viners encouraged the team to hold off.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

“It was actually the Viners who were like ‘no, no — this is way more important than just nostalgia,” he explains. The users said they wanted something like Vine that would reset social media and filter out AI slop. “They’re the ones who told us to wait and get it right. And so that’s what we did.”

The team rolled up their sleeves, revisited and rewrote the code in parts, and focused on the app’s design.

The version of Divine launching today is a result of those efforts, and includes features like compilation mode, which reflects how much of today’s youngest generation grew up watching Vines. With this mode, people can create their own lists of videos to make their own compilations.

To use compilations, you could visit a hashtag, like #cats, and it will autoplay a stream of matching Vine videos. You can stop the stream to interact with the content, like reposting or liking the videos, or you can just sit back and watch.

Image Credits:Divine

Another key aspect of Divine’s value proposition is its lack of AI-generated content.

“I decided that I was going to filter out AI content because I personally don’t like seeing AI content. I don’t like feeling tricked,” Rabble says. “I don’t like the idea that tons of content can be made very quickly and with little humanity or thought.”

To keep the network free from this unwanted content, or “AI slop,” as it’s often called, DiVine requires users to either record videos directly in the app or verify how uploaded videos were created using C2PA, an open industry standard that establishes the origin and the edits to digital content.

Beyond offering the experience of bringing back Vine, the app’s other mission is to popularize open protocols — something that delivers on Rabble’s vision for reclaiming social media from the tech giants. Divine itself is built on the open social protocol Nostr, and the team is experimenting with integrating the open source AT Protocol, which powers Bluesky. In the future, Divine may also integrate with the ActivityPub protocol, which underpins alternative social networks like Mastodon and Flipboard, and is built into Meta’s Threads.

Divine has no revenue model and is structured as a public benefit corporation. But Rabble believes that it could allow digital creators to take back some control over their online presence, which they could monetize via brand deals or collabs, as they do currently. He also likes the Patreon model for supporting creators directly and the idea of a Pro account that would provide additional features.

“Many of us came from Vine, and it was the beginning of everything,” said OG Viner, Lele Pons, about the app’s relaunch. “An iconic app. It was such a key moment in my own personal journey, and in internet culture, it makes me so happy to see these early classics brought back to life, and to have the chance to make new ones.”

Divine is available as a free download on the App Store, Google Play, and the Nostr-powered Zapstore. Initially, it will roll out access to those on the waitlist and will let others in gradually via the use of invite codes.

At a U.S. military base in central California, four-seater all-terrain vehicles roam hillside trails. This is a training exercise, but not for the people in the vehicles: It’s an effort to train AI models to enter conflict zones.

The autonomous military ATVs are operated by Scout AI, a startup founded in 2024 by Coby Adcock and Collin Otis, that calls itself a “frontier lab for defense.” The company said on Wednesday that it has raised $100 million in a Series A round led by Align Ventures and Draper Associates, following its $15 million seed round in January 2025.

Scout invited TechCrunch for an exclusive tour of its training operations at a military base it asked us not to name.

The company is building an AI model it calls “Fury” to operate and command military assets, first for logistical support, but soon for autonomous weapons. CTO Collin Otis compares this work, which builds on existing LLMs, to training soldiers.

“[Soldiers] start when they’re 18 years old, and sometimes they even start after college, so you want to start with that base level of intelligence,” Otis told TechCrunch. “It’s useful to start with someone who’s already made an investment and then say, ‘Hey, what do I have to do to teach this thing to be an incredible military AGI, versus just being a broadly intelligent AGI?’”

Scout has secured military technology development contracts totaling $11 million from organizations like DARPA, the Army Applications Laboratory, and other Department of Defense customers. It is one of 20 autonomy companies whose technology is being used by the U.S. Army’s 1st Cavalry Division during its regular training cycle at Fort Hood in Texas, with the expectation that the unit will bring along products that prove themselves when it next deploys in 2027.

For Scout’s internal testing, the rubber meets the dirt in the base’s hilly terrain, where the company’s operations team, led by former soldiers, is putting the vehicles through their paces on simulated missions.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

Autonomous cars are starting to be seen in more cities around the world, but they operate in more structured environments with rules. Operating autonomously on unmarked trails or off-road is another challenge entirely. Otis, who previously worked at autonomous trucking company Kodiak, said he was motivated to start Scout when he realized the system he helped build there wasn’t intelligent enough to operate in an unpredictable war zone.

An autonomous ground vehicle controlled by Scout AI’s Fury model. Image Credits:Scout Ai / Scout AI

A new approach to autonomy

Scout is turning to a newer autonomy technology: Vision Language Action models, or VLAs, that are based on LLMs and used to control robots. First released by Google DeepMind in 2023, the technology seeded robotics startups like Physical Intelligence and Figure.AI, the humanoid robot company led by Adcock’s brother, Brett.

Coby Adcock is on Figure’s board, and he says that experience convinced him of the opportunity to bring broader intelligence to the military’s growing fleet of autonomous vehicles. His brother introduced him to Otis, who was advising Figure, and they set about applying the latest in AI to military solutions.

“If I handed you the controller of a drone right now and I strapped a headset on you, you could learn to fly that thing in minutes,” Otis said. “You’re actually just learning how to connect your prior knowledge to these couple little joysticks. It’s not a big leap. That’s the way to think about VLAs and why they’re such an unlock.”

Indeed, I got a chance to drive one of Scout’s ATVs around the rutty trails, and the terrain was challenging: steep hills, loose sand on turns, disappearing tracks, confusing intersections. I’m not an experienced ATV driver, but made a fair go on my first attempt (if I say so myself). That’s the kind of general intelligence the company wants in its models, which it has been training using these ATVs for just six weeks — it started off using civilian ATVs.

I also rode in the ATV under autonomous control, and could feel the difference — it accelerates faster than a human who might be thinking about a passenger’s comfort. The operations team pointed out how the vehicles hug the right on wider trails, but stay in the middle of narrow ones, like their training drivers. They also, when confused, suddenly slow down to think over their next move, which happened a few times as the ATV carried us on a 6.5 km loop before returning to base.

Though the VLAs are new enough that they are yet to be deployed by any company in an operational setting, “the technology is good enough to be doing that experimentation in the field with soldiers to figure out how to most be effective to U.S. forces,” said Stuart Young, a former DARPA program manager who worked on ground vehicle autonomy. And like other autonomy companies, Scout’s full stack also includes deterministic systems and other flavors of AI to round out its agents’ capabilities.

Young left DARPA this month to join Field after managing a program called RACER that asked companies to create high-speed, autonomous off-road vehicles, helping seed this space the same way that the organization’s Grand Challenge boosted self-driving cars. Two competitors in this space, Field AI and Overland AI, were spun out of that program, and Scout also participated later.

The first applications of ground autonomy, according to Scout executives and military technologists, will be automated resupply: Carrying water or ammunition to distant observation posts, or in a convoy where a crewed truck might be followed by six to ten autonomous vehicles, saving precious human labor for more important tasks.

Brian Mathwich, an active duty infantry officer doing a stint as a military fellow at Scout, recalled a recent exercise in Alaska where he led a resupply convoy in total darkness and wished for autonomous vehicles to help him out.

Image Credits:Scout AI / Scout AI

Adding intelligence to the Army’s motor pool

Scout sees itself primarily as a software company building an intelligence layer for military machines. It doesn’t intend to make the autonomous vehicles themselves, but instead build atop them.

Adcock expects the startup’s first product to be widely adopted will be one called “Ox,” a command and control software bundled with hardened computer hardware like GPUs, communications and cameras. It’s intended to allow individual soldiers to orchestrate multiple drones and autonomous ground vehicles using prompts such as, “Go to this waypoint and watch for enemy forces.”

However, making that software work requires training on real vehicles, which is why it has set up Foundry, its training range at the military base. There, drivers spend eight-hour shifts putting the ATVs through their paces, then work through a reinforcement learning system to log where they had to take over, which is used to improve the model. The base commander has even asked the company’s ATV to take a turn with security patrols.

One hypothesis Scout is testing is that VLAs will enable this relatively limited data set, alongside training data in simulations, to deliver a fully capable driving agent. While the vehicle seems comfortable on trails, for example, it isn’t ready to operate fully off-road.

Scout is also practicing with drones for reconnaissance and defense, giving them intelligence with vision language models.

The startup is working on a system that would see groups of munition drones fly with a larger “quarterback” platform that provides more compute resources to command them. For example, the drones could search a geographic area for hidden enemy tanks and attack them, possibly without human intervention. Otis contends that the alternative approach in such a scenario might be indirect artillery fire, which is imprecise compared to drone strikes.

While autonomous weapons are a flash point in the politics of defense tech, experts note the concept is old: Heat-seeking missiles and mines have been used in warfare for decades. The question for technologists is how the weapons are controlled, according to Jay Adams, a retired U.S. Army Captain who leads Scout’s operations team.

Adams notes the company’s munitions drones can be programmed to only attack threats in a specific geographic area, or only following human confirmation. He also says autonomous weapons platforms are unlikely to fire because they are scared, the way an 18-year-old soldier might be.

VLAs, too, hold promise for improving targeting. Scout says its models are pre-trained on a specific set of military data to prepare them for, say, running into an enemy tank while on a resupply mission. Lt. Col. Nick Rinaldi, who supervises Scout’s work for the Army Applications Laboratory, says that while automated targeting is hard and unlikely to be used outside of constrained environments in the near term, the potential of VLAs to reason about threats make them a promising technology to investigate.

Adams says the promise of drones that can identify their own targets is key to warfare in the future. While Russia’s invasion of Ukraine has generated intense interest in drone warfare, he believes that humans operating individual UAVs doesn’t scale well enough for the U.S. to face a large number of low-cost unmanned systems should they threaten U.S. forces.

A mission to counter anti-military vibes

Image Credits:Scout AI / Scout AI

Like many defense startups, Scout wears its mission on its sleeve, and its executives will freely criticize companies that are reluctant to hand their technology over to the government. Google, for example, reportedly pulled out of a Pentagon contest to develop control systems for autonomous drone swarms, a capability Scout is also working on.

“The AI people don’t want to work with the military,” Otis told TechCrunch, referencing Anthropic’s spat with the Pentagon over its terms of service. “None of them are open to running agents on one-way attack drones, or running agents on missile systems.”

Nevertheless, Scout is using existing LLMs as the base to build its agents, though it declined to say which ones. Otis says it has agreements with “very well-known hyperscalers” to provide the pre-trained intelligence for Scout’s foundation model. He also declined to say if the company uses open-weight models, such as those offered by Chinese companies. Many companies reliant on AI inference build on top of open-weight models because they are cheaper compared to offerings from frontier labs like Anthropic or OpenAI.

Scout expects to address this by building its own model from the ground up in the years ahead, and the founders say much of its capital will go into those training and compute costs. Indeed, Otis wonders if Scout will beat the existing leaders to AGI because its model will be constantly interacting with the real world.

“There’s an argument in the AGI community along the lines that you can only get so intelligent by reading the internet, and most intelligence comes with interacting in the world,” Otis said.

Does that mean Adcock is competing with his brother’s army of humanoid robots at Figure? No, Otis says, but “we can get to scale much faster because our customer has assets,” he said, referring to the Pentagon.

At a U.S. military base in central California, four-seater all-terrain vehicles roam hillside trails. This is a training exercise, but not for the people in the vehicles: It’s an effort to train AI models to enter conflict zones.

The autonomous military ATVs are operated by Scout AI, a startup founded in 2024 by Coby Adcock and Collin Otis, that calls itself a “frontier lab for defense.” The company said on Wednesday that it has raised $100 million in a Series A round led by Align Ventures and Draper Associates, following its $15 million seed round in January 2025.

Scout invited TechCrunch for an exclusive tour of its training operations at a military base it asked us not to name.

The company is building an AI model it calls “Fury” to operate and command military assets, first for logistical support, but soon for autonomous weapons. CTO Collin Otis compares this work, which builds on existing LLMs, to training soldiers.

“[Soldiers] start when they’re 18 years old, and sometimes they even start after college, so you want to start with that base level of intelligence,” Otis told TechCrunch. “It’s useful to start with someone who’s already made an investment and then say, ‘Hey, what do I have to do to teach this thing to be an incredible military AGI, versus just being a broadly intelligent AGI?’”

Scout has secured military technology development contracts totaling $11 million from organizations like DARPA, the Army Applications Laboratory, and other Department of Defense customers. It is one of 20 autonomy companies whose technology is being used by the U.S. Army’s 1st Cavalry Division during its regular training cycle at Fort Hood in Texas, with the expectation that the unit will bring along products that prove themselves when it next deploys in 2027.

For Scout’s internal testing, the rubber meets the dirt at in the base’s hilly terrain, where the company’s operations team, led by former soldiers, is putting the vehicles through their paces on simulated missions.

Techcrunch event Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. Meet your next investor or portfolio startup at Disrupt

Your next round. Your next hire. Your next breakout opportunity. Find it at TechCrunch Disrupt 2026, where 10,000+ founders, investors, and tech leaders gather for three days of 250+ tactical sessions, powerful introductions, and market-defining innovation. Register now to save up to $410. San Francisco, CA | REGISTER NOW

Autonomous cars are starting to be seen in more cities around the world, but they operate in more structured environments with rules. Operating autonomously on unmarked trails or off-road is another challenge entirely. Otis, who previously worked at autonomous trucking company Kodiak, said he was motivated to start Scout when he realized the system he helped build there wasn’t intelligent enough to operate in an unpredictable war zone.

An autonomous ground vehicle controlled by Scout AI’s Fury model. Image Credits:Scout Ai / Scout AI

A new approach to autonomy

Scout is turning to a newer autonomy technology: Vision Language Action models, or VLAs, that are based on LLMs and used to control robots. First released by Google DeepMind in 2023, the technology seeded robotics startups like Physical Intelligence and Figure.AI, the humanoid robot company led by Adcock’s brother, Brett.

Coby Adcock is on Figure’s board, and he says that experience convinced him of the opportunity to bring broader intelligence to the military’s growing fleet of autonomous vehicles. His brother introduced him to Otis, who was advising Figure, and they set about applying the latest in AI to military solutions.

“If I handed you the controller of a drone right now and I strapped a headset on you, you could learn to fly that thing in minutes,” Otis said. “You’re actually just learning how to connect your prior knowledge to these couple little joysticks. It’s not a big leap. That’s the way to think about VLAs and why they’re such an unlock.”

Indeed, I got a chance to drive one of Scout’s ATV around the rutty trails, and the terrain was challenging: steep hills, loose sand on turns, disappearing tracks, confusing intersections. I’m not an experienced ATV driver, but made a fair go on my first attempt (if I say so myself). That’s the kind of general intelligence the company wants in its models, which it has been training using these ATVs for just six weeks — it started off using civilian ATVs.

I also rode in the ATV under autonomous control, and could feel the difference — it accelerates faster than a human who might be thinking about a passenger’s comfort. The operations team pointed out how the vehicles hug the right on wider trails, but stay in the middle of narrow ones, like their training drivers. They also, when confused, suddenly slow down to think over their next move, which happened a few times as the ATV carried us on a 6.5 km loop before returning to base.

Though the VLAs are new enough that they are yet to be deployed by any company in an operational setting, “the technology is good enough to be doing that experimentation in the field with soldiers to figure out how to most be effective to U.S. forces,” said Stuart Young, a former DARPA program manager who worked on ground vehicle autonomy. And like other autonomy companies, Scout’s full stack also includes deterministic systems and other flavors of AI to round out its agents’ capabilities.

Young left DARPA this month to join Field after managing a program called RACER that asked companies to create high-speed, autonomous off-road vehicles, helping seed this space the same way that the organization’s Grand Challenge boosted self-driving cars. Two competitors in this space, Field AI and Overland AI, were spun out of that program, and Scout also participated later.

The first applications of ground autonomy, according to Scout executives and military technologists, will be automated resupply: Carrying water or ammunition to distant observation posts, or in a convoy where a crewed truck might be followed by six to ten autonomous vehicles, saving precious human labor for more important tasks.

Brian Mathwich, an active duty infantry officer doing a stint as a military fellow at Scout, recalled a recent exercise in Alaska where he led a resupply convoy in total darkness and wished for autonomous vehicles to help him out.

Image Credits:Scout AI / Scout AI

Adding intelligence to the Army’s motor pool

Scout sees itself primarily as a software company building an intelligence layer for military machines. It doesn’t intend to make the autonomous vehicles themselves, but instead build atop them.

Adcock expects the startup’s first product to be widely adopted will be one called “Ox,” a command and control software bundled with hardened computer hardware like GPUs, communications and cameras. It’s intended to allow individual soldiers to orchestrate multiple drones and autonomous ground vehicles using prompts such as, “Go to this waypoint and watch for enemy forces.”

However, making that software work requires training on real vehicles, which is why it has set up Foundry, its training range at the military base. There, drivers spend eight-hour shifts putting the ATVs through their paces, then work through a reinforcement learning system to log where they had to take over, which is used to improve the model. The base commander has even asked the company’s ATV to take a turn with security patrols.

One hypothesis Scout is testing is that VLAs will enable this relatively limited data set, alongside training data in simulations, to deliver a fully capable driving agent. While the vehicle seems comfortable on trails, for example, it isn’t ready to operate fully off-road.

Scout is also practicing with drones for reconnaissance and defense, giving them intelligence with vision language models.

The startup is working on a system that would see groups of munition drones fly with a larger “quarterback” platform that provides more compute resources to command them. For example, the drones could search a geographic area for hidden enemy tanks and attack them, possibly without human intervention. Otis contends that the alternative approach in this scenario might be indirect artillery fire, which is imprecise compared to drone strikes.

While autonomous weapons are a flash point in the politics of defense tech, experts note the concept is old: Heat-seeking missiles and mines have been in use for many decades. The question for technologists is how the weapons are controlled, Jay Adams, a retired U.S. Army Captain who leads Scout’s operations team, told TechCrunch.

He notes the company’s munitions drones can be programmed to only attack threats in a specific geographic area, or only with human confirmation. He also says autonomous weapons platforms are unlikely to fire because they are scared, the way an eighteen year-old soldier might.

VLAs, too, offer promise for better targeting. Scout says its models are pretrained on a specific set of military data to prepare them for, say, running into an enemy tank while on a resupply mission. Lt. Col Nick Rinaldi, who supervises Scout’s work for the Army Applications Laboratory, says that while automated targeting is hard and unlikely to be used outside of constrained environments in the near term, the potential of VLAs to reason about threats make them a promising technology to investigate.

Adams says the promise of drones that can identify their own targets is key to future warfare: While Russia’s invasion of Ukraine has generated intense interest in drone warfare, he believes having humans operating individual UAVs doesn’t scale enough for the US to face a large number of low-cost unmanned systems should they threaten US forces.

A mission to counter anti-military vibes

Image Credits:Scout AI / Scout AI

Like many defense startups, Scout wears its mission on its sleeve, and executives will freely criticize companies that are reluctant to hand their technology over to the government. Google, for example, reportedly pulled out of a Pentagon contest to develop control systems for autonomous drone swarms, a capability Scout is also working on.

“The AI people don’t want to work with the military,” Otis told TechCrunch, referencing Anthropic’s spat with the Pentagon over its terms of service. “None of them are open to running agents on one-way attack drones, or running agents on missile systems.”

Nevertheless, Scout is actually using existing LLMs as the base to build its agents, though declined to say which ones. Otis says it has agreements with “very well known hyperscalers” to provide the pretrained intelligence for Scout’s foundation model. Otis also declined to comment on if it uses open-weight models, such as those offered by Chinese companies. Many companies reliant on AI inference build on these models to operate with lower cost compared to models from frontier labs like Anthropic or OpenAI.

Scout expects to address this by building its own model from the ground up in the years ahead, and the founders say much of its capital will go into those training and compute costs. Indeed, Otis wonders if Scout will beat the existing leaders to AGI because its model will be constantly interacting with the real world.

“There’s an argument in the AGI community along the lines that you can only get so intelligent by reading the internet, and most intelligence comes with interacting in the world,” Otis said.

Does that mean Adcock is competing with his brother’s army of humanoid robots at Figure? No, Otis says, but “we can get to scale much faster because our customer has assets,” he said, referring to the Pentagon.

專攻AI大數據分析的老牌大廠SAS今年迎來50周年，從北卡羅來納州立大學的農業數據研究起家，到如今服務逾150個國家、多數為財富500企業，他們在年度大會首日（4/28）上重申AI戰略方向，聚焦可信任AI，要從決策流程、治理機制和資料等三層面下手，打造企業級可信任且可規模化的AI應用架構。他們也以一系列產品布局來支持這個戰略方向，包括以MCP結合大型語言模型的決策平臺、這次新推出的AI治理工具SAS AI Navigator，以及資料管理和合成資料工具。

代理型AI帶來決策可信任挑戰

原本只是要解決農業運算需求的SAS，在1976年發表同名程式語言，1998年開始在服務中融入機器學習功能、2016年納入深度學習、2020年擁抱多雲架構，確保自家軟體服務能在多雲環境和地端執行。2024年，他們還推出合成資料生成器，來供資料不足、但想訓練AI模型的企業使用。

SAS技術長Bryan Harris在大會上指出，他們這一路發展降低了企業用AI的門檻，卻也發現可信任問題，尤其是代理型AI（Agentic AI）應用，「所有AI代理工作流程，都包含了確定性（Deterministic）與非確定性（Non-deterministic）元素。」

其中，確定性系統可提供穩定且可預測的結果，但以大型語言模型（LLM）為基礎的生成式AI應用，可能因上下文不同而產生不同輸出，進而影響企業決策品質。

解方是LLM當介面、確定性邏輯做為決策關鍵

Bryan Harris以反洗錢警示為例，進一步說明為何不能把所有決策都交給LLM。同一筆可疑交易，LLM跑三次可能得出三種不同結論：低風險、高風險、或直接關閉警示。如果企業把這種判斷完全交給LLM，出錯的風險會以複合方式累積。

SAS的解法是，結合確定性和非確定性邏輯，讓LLM當作前臺介面，來處理語言理解、推理和解釋，而SAS的規則引擎處理則作為後臺決策核心，確保關鍵決策都在可追溯、可稽核的確定性系統上執行。

SAS全球AI暨生成式AI產品策略負責人Marinela Profi以實例說明，一家零售業者店長收到庫存預警後，直接在Claude的對話介面問「我該怎麼辦？」Claude呼叫了SAS Intelligent Decisioning智慧決策平臺，根據銷售預測、庫存數據和補貨政策，算出「建議補貨約1,040件」，還附上詳細的決策依據和稽核紀錄。

在整個過程中，使用者都透過Claude這款LLM介面來對話與解釋，SAS的決策平臺則負責實際運算和政策套用，兩者以MCP伺服器串接，解決決策可信任問題。

影子AI問題靠禁止沒用，要讓治理變得夠方便才有效

同樣是要打造可信任AI系統，SAS在今年大會上揭露了新治理工具AI Navigator，是整個戰略的一大亮點產品。它的核心邏輯很簡單：與其等AI出問題再來收拾，不如一開始就把治理嵌進去，尤其是在員工使用未經授權的AI來解決公司問題、影子AI（Shadow AI）風險越來越高的現在。

SAS副總裁Reggie Townsend負責帶領SAS數據倫理措施（Data Ethics Practice）團隊，他在現場展示，當企業使用者發現某個開源LLM被標記為資料投毒（Data poisoning）風險，AI Navigator五分鐘內就能完成這些工作：確認這個模型連結到哪些應用場景、套用了哪些法規政策、找到合規團隊已有的應變措施、發現一個可以替代的授權代理，並通知資產負責人啟動處理流程。

他直言，AI治理計畫最大的風險，不是法規，「而是工具太複雜、沒人真的去用。」他也點出，企業導入AI面臨的3大核心張力，比如在道德層面，AI必須符合企業的價值觀與信念；在營運層面，效率提升不能犧牲一致性和清晰度；在財務層面，要創造AI價值但不能提高失控風險。

他強調，這三者都需要靠「好的判斷力」來平衡，而AI治理正是把這種判斷力系統化、規模化的機制。

以數位分身和合成資料補足現實限制，讓What if變成可模擬的問題

在應用層面，SAS也持續推動數位分身（Digital Twin），作為驗證和優化企業決策流程的核心工具。

Bryan Harris指出，去年，他們就與Epic Games合作，在Unreal Engine中建置光擬真的虛擬場域，來讓企業在不冒真實風險的情況下，模擬各種營運情境。

他在現場舉例，丹麥最大醫療器材滅菌廠商SteriCentral，就用這套數位分身服務來解決問題。他們原本以為，醫療器械清洗流程的低效率，來自緩衝升降機卡滯；但導入數位分身模擬後才發現，升降機本身就是整個流程的中央分流節點，所有器械托盤都要經過它重新分配，才導致排隊延遲。

此為數位分身模擬的醫療器械清洗流程。

釐清癥結後，團隊調整分流策略，成功化解瓶頸、提高作業效率。甚至更進一步，SAS還用同一套數位分身工具產出合成影像，來訓練電腦視覺模型，偵測廠商人員作業時，是否正確佩戴口罩、手套等防護裝備。

Bryan Harris也透露其他產業的實例。比如能源業方面，SAS正訓練無人機搭配電腦視覺，提前偵測電力設備故障，預防停電和可能造成的野火；油氣業則建置深海鑽油平臺的數位分身，能即時偵測氣體洩漏等危險情況。Bryan Harris稱數位分身為「終極的What-if實驗場」，讓企業在真正行動前，就能看清楚每個決策的可能後果。

遷移不一定會痛苦：Posten Bring的雲端升級經驗

除了主打AI可信任機制，SAS還在大會首日請來挪威郵政物流集團Posten Bring，不只分享他們從SAS 9.4升級到SAS Viya的經驗，還包括後來使用的新款資料分析工具好處。

Posten Bring分析部門主管Bendik Overgaard說明了遷移的三個主要原因：公司原本就有全面上雲的技術策略；24小時不間斷的物流營運需要即時資料支援，但舊平臺容量已經跟不上需求；SAS提供的遷移工具可讓超過95%的既有程式碼直接沿用，大幅降低重建成本。

於是，雙方展開遷移，完成後全集團6、700名主管，每天上班都可以即時獲得所需的管理資訊。不只如此，他們還採用新款資料分析工具SpeedyStore，直接在資料所在處就能執行分析，不需要把資料搬移到特定平臺，不只分析速度更快，資料主權也更有保障。

遷移完成後，全集團6、700名主管每天上班，都能即時獲得所需的管理資訊。這背後靠的是SAS SpeedyStore資料分析工具，過去企業做分析，習慣把資料搬到分析平臺再處理，但SpeedyStore反過來，讓分析運算直接在資料所在處執行，資料不動、分析移過去。這樣做不只速度更快、成本更低，資料不離開原本的環境，資安風險也跟著降低。

開源終端機模擬器Ghostty的作者，同時也是HashiCorp共同創辦人的Mitchell Hashimoto，於4月28日在個人部落格宣布，將把Ghostty專案搬離使用超過18年的GitHub。他提到，問題不在Git本身，而在GitHub的Issues、拉取請求（Pull Requests，PR）與Actions等周邊服務近月頻繁故障，已嚴重影響其日常開發。新託管平臺尚未公布，Mitchell Hashimoto正與多家商業與開源方案洽談，現有儲存庫將保留為唯讀鏡像，URL維持不變。

Mitchell Hashimoto為GitHub第1,299號用戶，他表示，自2008年2月註冊以來，幾乎每天都會開啟GitHub。他曾共同創辦HashiCorp，並打造Vagrant、Terraform與Vault等基礎設施工具，目前主力放在2021年起開發的Ghostty，這是一款採用GPU加速與平臺原生UI的跨平臺終端機模擬器。

Mitchell Hashimoto提到，過去一個月，他以日記方式逐日記錄受GitHub故障影響的工作日，幾乎每天都被畫上記號。他撰寫該部落格文章當下，還因GitHub Actions故障，無法進行Pull Request審查長達約兩小時，他特別強調，問題不在Git本身，而是GitHub周邊基礎設施，包括Issues、PR與Actions等服務。

事實上，GitHub在4月下旬接連發生多起事故，4月24日Pull Requests服務發生影響合併佇列操作的迴歸問題，使用擠壓合併（Squash Merge）且合併群組包含多個Pull Request時，會產生錯誤的合併提交，期間影響658個儲存庫與2,092個Pull Request。

另一起搜尋相關事故則發生於4月28日，GitHub表示Elasticsearch叢集過載，可能與殭屍網路攻擊有關，導致Issues、Pull Requests、Projects與Actions工作流程執行頁面等仰賴搜尋的功能出現異常。

關於搬遷規畫，Mitchell Hashimoto表示會以漸進方式移除Ghostty對GitHub的依賴，未來數月再公布更多細節。本次搬遷僅針對Ghostty，他其餘個人專案目前仍留在GitHub，他也表示，若GitHub未來在可靠度上有實際改善，不排除重返。

GitHub技術長Vlad Fedorov也在4月28日於官方部落格發文提到，GitHub自2025年10月起執行10倍容量擴充計畫，而到2026年2月，GitHub判斷未來需要支援目前30倍規模。Vlad Fedorov將主因歸於代理式開發流程快速增加，帶動儲存庫建立、PR活動、API使用、自動化與大型儲存庫工作負載成長。