VentureBeat

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More GitHub helped to kick off the modern era of using AI to build applications with its Copilot technology and now it’s looking to open AI up even more. At the GitHub Universe conference today, the company rolled out an expansion of its AI-powered development tools. To date, GitHub Copilot has relied on OpenAI’s large language models (LLMs), including OpenAI Cortex in the beginning, to power its technology. Now GitHub is going multi-model. GitHub Copilot now supports multiple AI models, allowing developers to choose between Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro and OpenAI’s GPT4o variants. The GitHub Models service which was first announced in August is also growing, providing users with more ways and options to try out LLMs in a model playground. There is now even more integration with Microsoft’s VS Code integrated development environment (IDE), that enables multi-file editing. Agentic AI is also getting a boost with a series of updates to the GitHub Copilot Workspace service. Going a step further, the new GitHub Spark technology is an attempt to make it even easier to build basic applications quickly in an effort to enable more people to develop applications.  Rounding out the GitHub Universe update is an expansion of Copilot to support the Apple Xcode IDE and the availability of a StackOverflow extension  “We’re taking the Copilot platform from single-threaded to multi-threaded,” Mario Rodriguez, Chief Product Officer at GitHub told VentureBeat. What multi-model AI means for GitHub Copilot users Expanding the available AI models for use with GitHub Copilot provides numerous benefits to enterprises and their developers. Rodriguez noted that now users will have the ability to choose from different AI models to accomplish their coding tasks, rather than being limited to a single model. He said that just like there is more than one programming language, there are many LLMs to choose from and each has its own benefits. At launch, developers will still have to choose if they want to use a different model than OpenAI. Rodriguez said that in the future, Copilot may be able to automatically select the most appropriate model for a given task, based on factors like speed and performance, to provide the best results. Enhanced code editing and review land in the GitHub universe GitHub is introducing significant improvements to its VS Code integration, including multi-file editing capabilities. The new feature allows developers to instruct Copilot to make changes across multiple files simultaneously, rather than editing each file individually. A new code review system, currently in private preview and moving to public preview, has received very positive feedback according to GitHub. The system allows teams to configure specific rules and requirements, with Copilot automatically reviewing pull requests based on team-level configurations. “Code review is the essence of iteration velocity,” Rodriguez noted. “If you’re a developer, and you finish some code, and you have it in code review, and you’re waiting and waiting and waiting for feedback… that’s code sitting there that is not in production. The faster you can get feedback, the better it is.” GitHub Copilot comes to Apple Xcode  GitHub is also expanding the reach of Copilot with a series of new options. While GitHub Copilot has always been integrated with Microsoft’s VS code IDE, it wasn’t available for users of Apple’s Xcode. That’s no longer the case. “We want Copilot to be everywhere,” Rodriguez said. “So we already have it in JetBrains, in the terminal  and now it’s in Xcode.” Stack Overflow and GitHub partnership expands with new extensions Beyond just being available in other developer tools, GitHub wants to be an integrated part of the larger development ecosystem. A core part of that ecosystem in recent years is the StackOverflow community, where developers ask questions and share tips on development practices. At GitHub Universe, Stack Overflow announced the availability of its GitHub Copilot Extension. The new extension allows developers to get insight from Stack Overflow directly within GitHub Copilot. Prashanth Chandrasekar, CEO of Stack Overflow, told VentureBeat that AI can help developers work faster, eliminating cycles and freeing up headspace for higher-level work.  “However, one key caveat to keep in mind: AI can generate code, but it can’t provide the context, history or background on whether that code will fit the need and work as the question asked,” Chandrasekar said. “Our hope is that this extension will be used in a way to help support those looking for highly technical, trusted knowledge with the sources cited to back up what the user is looking for.” Agentic AI advances with GitHub Copilot Workspaces GitHub’s Workspace feature, which has already attracted more than 100,000 developers in preview, is receiving significant updates.  The platform now offers enhanced integration with GitHub.com, including a new pull request experience that allows developers to quickly address code suggestions and resolve issues through an AI-native interface. Rodriguez explained that the system acts as an orchestration engine, similar to how Kubernetes orchestrates infrastructure for the cloud, but for AI-powered development tools. This allows developers to move seamlessly from idea to implementation using natural language interactions. GitHub lights a new Spark for software creation Perhaps the most ambitious announcement is Spark, a new tool aimed at making software development accessible to non-professionals. The platform allows users to quickly create personal applications without extensive coding knowledge. Unlike traditional low-code or no-code platforms, Spark focuses on enabling personal software creation for joy and creativity. Spark is using the power of Copilot to create the applications. Rodriguez demonstrated this by sharing how he created a math game for his daughter in just five minutes, emphasizing the platform’s accessibility and immediate utility. “The goal is 1 billion developers,” Rodriguez explained. “By 2030 we might have 10 billion people in the world, wouldn’t it be amazing if we could actually unlock the power of creating software for 1 billion of them?” source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More AI video has come incredibly far in the years since the first models debuted in late 2022, increasing in realism, resolution, fidelity, prompt adherence (how well they match the text prompt or description of the video that the user typed) and number. But one area that remains a limitation to many AI video creators — myself included — is in depicting realistic facial expressions in AI generated characters. Most appear quite limited and difficult to control. But no longer: today, Runway, the New York City-headquartered AI startup backed by Google and others, announced a new feature “Act-One,” that allows users to record video of themselves or actors from any video camera — even the one on a smartphone — and then transfers the subject’s facial expressions to that of an AI generated character with uncanny accuracy. The free-to-use tool is gradually rolling out “gradually” to users starting today, according to Runway’s blog post on the feature. While anyone with a Runway account can access it, it will be limited to those who have enough credits to generate new videos on the company’s Gen-3 Alpha video generation model introduced earlier this year, which supports text-to-video, image-to-video, and video-to-video AI creation pipelines (e.g. the user can type in a scene description, upload an image or a video, or use a combination of these inputs and Gen-3 Alpha will use what its given to guide its generation of a new scene). Despite limited availability right now at the time of this posting, the burgeoning scene of AI video creators online is already applauding the new feature. As Allen T. remarked on his X account “This is a game changer!” It also comes on the heels of Runway’s move into Hollywood film production last month, when it announced it had inked a deal with Lionsgate, the studio behind the John Wick and Hunger Games movie franchises, to create a custom AI video generation model based on the studio’s catalog of more than 20,000 titles. Simplifying a traditionally complex and equipment-heavy creative proccess Traditionally, facial animation requires extensive and often cumbersome processes, including motion capture equipment, manual face rigging, and multiple reference footages. Anyone interested in filmmaking has likely caught sight of some of the intricacy and difficulty of this process to date on set or when viewing behind the scenes footage of effects-heavy and motion-capture films such as The Lord of the Rings series, Avatar, or Rise of the Planet of the Apes, wherein actors are seen covered in ping pong ball markers and their faces dotted with marker and blocked by head-mounted apparatuses. Accurately modeling intricate facial expressions is what led David Fincher and his production team on The Curious Case of Benjamin Button to develop whole new 3D modeling processes and ultimately won them an Academy Award, as reported in a prior VentureBeat report. Yet in the last few years, new software and AI-based startups such as Move have sought to reduce the equipment necessary to perform accurate motion capture — though that company in particular has concentrated primarily on full-body, more broad movements, whereas Runway’s Act-One is focused more on modeling facial expressions. With Act-One, Runway aims to make this complex process far more accessible. The new tool allows creators to animate characters in a variety of styles and designs, without the need for motion-capture gear or character rigging. Instead, users can rely on a simple driving video to transpose performances—including eye-lines, micro-expressions, and nuanced pacing—onto a generated character, or even multiple characters in different styles. As Runway wrote on its X account: “Act-One is able to translate the performance from a single input video across countless different character designs and in many different styles.” The feature is focused “mostly” on the face “for now,” according to Cristóbal Valenzuela, co-founder and CEO of Runway, who responded to VentureBeat’s questions via direct message on X. Runway’s approach offers significant advantages for animators, game developers, and filmmakers alike. The model accurately captures the depth of an actor’s performance while remaining versatile across different character designs and proportions. This opens up exciting possibilities for creating unique characters that express genuine emotion and personality. Cinematic realism across camera angles One of Act-One’s key strengths lies in its ability to deliver cinematic-quality, realistic outputs from various camera angles and focal lengths. This flexibility enhances creators’ ability to tell emotionally resonant stories through character performances that were previously hard to achieve without expensive equipment and multi-step workflows. The tool’s ability to faithfully capture the emotional depth and performance style of an actor, even in complex scenes. This shift allows creators to bring their characters to life in new ways, unlocking the potential for richer storytelling across both live-action and animated formats. While Runway previously supported video-to-video AI conversion as previously mentioned in this piece, which did allow users to upload footage of themselves and have Gen-3 Alpha or other prior Runway AI video models such as Gen-2 “reskin” them with AI effects, the new Act-One feature is optimized for facial mapping and effects. As Valenzuela told VentureBeat via DM on X: “The consistency and performance is unmatched with Act-One.” Enabling more expansive video storytelling A single actor, using only a consumer-grade camera, can now perform multiple characters, with the model generating distinct outputs for each. This capability is poised to transform narrative content creation, particularly in indie film production and digital media, where high-end production resources are often limited. In a public post on X, Valenzuela noted a shift in how the industry approaches generative models. “We are now beyond the threshold of asking ourselves if generative models can generate consistent videos. A good model is now the new baseline. The difference lies in what you do with the model—how you think about its applications and use cases, and what you ultimately build,” Valenzuela wrote. Safety and protection for public figure impersonations As with all of Runway’s

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More I’m a writer and always have been. My writing skills are undeniably central to my career as an in-house public relations leader and communications strategist. Admittedly, I scoffed at the notion of generative AI coming for my job. How could a soulless machine match my creative prowess? Eventually, I realized the threat to my career did not come from AI, but from my reluctance to adopt it.  Like many, I’ve been working for AI companies for years. I’ve worked with dozens of  AI-based applications, long before OpenAI’s launch of ChatGPT in November 2022 sent the world into a frenzy of fear and excitement.  Recently, at a marketing all-hands meeting, we were asked how often we use gen AI in our work. Everyone replied they were using it literally every day — except for me. There are times when you want to stand out in a crowd. This was not one of them. I suddenly felt like that uncle who still refuses to get a smartphone.  Letting go of pretentious skepticism I approached my first conscious encounter with a large language model (LLM) with a mixture of condescension and fear. Surely, no machine could replicate my professional wit and the tailored nuance of my prose, meticulously crafted and fit for purpose. It was an affront to my expertise and my pride to think I needed help from anyone or anything to do my best work. I also didn’t want to be seen as cutting corners.  Was using AI like cheating?  I quickly thought back to the impact of my writing on the trajectory of my life. Would I have gotten into Cornell if everyone was using AI to craft brilliant college essays? Has one of my greatest professional strengths now been democratized, chopped up into little easily accessible pieces and distributed to the masses? It felt like a talent I’ve cultivated for years was now everyone’s to tap into with just one click.  Existential dread popped in and out of my head. Was I a 2007 iPod?  Why was I so resistant to accepting AI into my work? It doesn’t take AI to figure out where my fear was coming from — a misconception that AI would replace me or, worse, make me average, rather than better. I saw AI-driven writing as a personal sleight, a harbinger signaling the redundancy of my craft. I was too afraid of the risk to my career to imagine the benefits. Falling for the enemy Faced with a growing to-do list and the new balancing act of returning from maternity leave to an expanded role leading public relations for a publicly-traded tech company, I opened Jasper AI.  I admittedly smirked at some of the functionality. Changing the tone? Is this AI emotionally intelligent? Maybe more so than some former colleagues. I began on a blank screen. I started writing a few lines and asked the AI to complete the piece for me. I reveled in the schadenfreude of its failure.  It summarized what I had written at the top of the document and just spit it out below. Ha! I had proven my superiority. I went back into my cave, denying myself and my organization the benefits of this transformative technology.  The next time I used gen AI, something in me changed. I realized how much prompting matters. You can’t just type a few initial sentences and expect the AI to understand what you want. It still can’t read our minds (I think). But there are dozens of templates that the AI understands. For PR professionals, there are templates for press releases, media pitches, crisis communications statements, press kits and more. And there are countless tools to discover. Prompting can be the difference between AI improving your writing and wasting a lot of time.  Models today can write coherent narratives, accurately use industry jargon, match tone and mirror any writing style. I would never copy and paste its work directly, because AI can infringe copyrights and hallucinate falsehoods, but it provides a great starting point and often conquers the initial “blank-page” battle of just sitting down and starting to write. Even just prompting the AI correctly forces you to bake out a decent outline, which is a great place to start for most writing projects. The impact on my time management and productivity was striking.  Using gen AI felt like I had the antidote to writer’s block.  I had found a first mate on my PR team who never takes days off.  Raising the bar Gen AI capabilities are making their way into countless business applications beyond writing-intensive professional domains like mine — and for good reason. Here’s my advice on making peace with these technologies:  No matter what you do for a living, stop swimming against the current. It will pull you under and take your career down with you. You need to ride this wave and master it.  Generative AI is not going to be your competitive advantage. Instead, it will likely raise the bar for everyone, moving the goalpost for your accomplishments whether you like it or not.  Don’t just regurgitate AI content. It’s obvious, detectable and doesn’t provide any value. Instead, thoughtfully harness gen AI to make what you’re already doing better, faster.  We don’t know how we’ll be using gen AI five years from now (or even next year), but rest assured, almost everyone reading this will be using it — whether they know it or not. AI will be integrated in holistic, human-centric and seamless ways across the apps we use at work and in daily life, as a vital part of systems we don’t see but that shapes our interactions.  As with so many things in life, adaptability and willingness to embrace change are the keys to staying relevant. Humans are resilient. AI isn’t coming for us. It’s coming for our inefficiencies. Grab these tools with both hands

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Very quickly, the topic of AI agents has moved from ambiguous concepts to reality. Enterprises will soon be able to deploy fleets of AI workers to automate and supplement — and yes, in some cases supplant — human talent.  “Autonomous agents are one of the hottest topics and perhaps one of the most hyped topics in gen AI today,” Gartner distinguished VP analyst Arun Chandrasekaran said at the Gartner Symposium/Xpo this past week.  However, while autonomous agents are trending on the consulting firm’s new generative AI hype cycle, he emphasized that “we’re in the super super early stage of agents. It’s one of the key research goals of AI companies and research labs in the long run.”  Top trends in Gartner’s AI Hype Cycle for gen AI Based on Gartner’s 2024 Hype Cycle for Generative AI, four key trends are emerging around gen AI — autonomous agents chief among them. Today’s conversational agents are advanced and versatile, but are “very passive systems” that need constant prompting and human intervention, Chandrasekaran noted. Agentic AI, by contrast, will only need high-level instruction that they can break out into a series of execution steps.  “For autonomous agents to flourish, models have to significantly evolve,” said Chandrasekaran. They need reasoning, memory and “the ability to remember and contextualize things.” Another key trend is multimodality, said Chandrasekaran. Many models began with text, and have since expanded into code, images (as both input and output) and video. A challenge in this is that “by the very aspect of getting multimodal, they’re also getting larger,” said Chandrasekaran.  Open-source AI is also on the rise. Chandrasekaran pointed out that the market has so far been dominated by closed-source models, but open source provides customization and deployment flexibility — models can run in the cloud, on-prem, at the edge or on mobile devices.  Finally, edge AI is coming to the fore. Much smaller models — between 1B to 10B parameters — will be used for resource-constrained environments. These can run on PCs or mobile devices, providing for “acceptable and reasonable accuracy,” said Chandrasekaran.  Models are “slimming down and extending from the cloud into other environments,” he said.  Heading for the trough At the same time, some enterprise leaders say AI hasn’t lived up to the hype. Gen AI is beginning to slide into the trough of disillusionment (when technology fails to meet expectations), said Chandrasekaran. But this is “inevitable in the near term.” There are a few fundamental reasons for this, he explained. First, VCs have funded “an enormous amount of startups” — but they have still grossly underestimated the amount of money startups need to be successful. Also, many startups have “very flimsy competitive moats,” essentially serving as a wrapper on top of a model that doesn’t offer much differentiation. Also, “the fight for talent is real” — consider the acqui-hiring models — and enterprises underestimate the amount of change management. Buyers are also increasingly raising questions about business value (and how to track it). There are also concerns about hallucination and explainability, and there’s more to be done to make models more reliable and predictable. “We are not living in a technology bubble today,” said Chandrasekaran. “The technologies are sufficiently advancing. But they’re not advancing fast enough to keep up with the lofty expectations enterprise leaders have today.” Not surprisingly, the cost of building and using AI is another significant hurdle. In a survey by Gartner, more than 90% of CIOS said that managing cost limits their ability to get value from AI. For instance, data preparation and inferencing costs are often greatly underestimated, explained Hung LeHong, a distinguished VP analyst at Gartner. Also, software vendors are raising their prices by up to 30% because AI is increasingly embedded into their product pipelines. “It’s not just the cost of AI, it’s the cost of applications they’re already running in their business,” said LeHong. Core AI use cases Still, enterprise leaders understand how instrumental AI will be going forward. Three-quarters of CEOs surveyed by Gartner say AI is the technology that will be most impactful to their industry, a significant leap from 21% just in 2023, LeHong pointed out.  That percentage has been “going up and up and up every year,” he said.  Right now, the focus is on internal customer service functions where humans are “still in the driver’s seat,” Chandrasekaran pointed out. “We’re not seeing a lot of customer-facing use cases yet with gen AI.”  LeHong pointed out that a significant amount of enterprise-gen AI initiatives are focused on augmenting employees to increase productivity. “They want to use gen AI at individual employee level.”  Chandrasekaran pointed to three business functions that stand out in adoption: IT, security and marketing. In IT, some uses for AI include code generation, analysis and documentation. In security, the technology can be used to augment SOCs when it comes to areas such as forecasting, incident and threat management and root cause analysis.  In marketing, meanwhile, AI can be used to provide sentiment analysis based on social media posts and to create more personalized content. “I think marketing and gen AI are made for each other,” said Chandrasekaran. “These models are quite creative.”  He pointed to some common use cases across these business functions: content creation and augmentation; data summarization and insights; process and workflow automation; forecasting and scenario planning; customer assistance; and software coding and co-pilots.   Also, enterprises want the ability to query and retrieve from their own data sources. “Enterprise search is an area where AI is going to have a significant impact,” said Chandrasekaran. “Everyone wants their own ChatGPT.”  AI is moving fast Additionally, Gartner forecasts that:  By 2025, 30% of enterprises will have implemented an AI-augmented and testing strategy, up from 5% in 2021.  By 2026, more than 100 million humans will engage with robo or synthetic virtual colleagues and nearly 80% of prompting will be semi-automated. “Models are going

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Universal Music Group (UMG), one of the largest recording labels in the world, has been vocal about the impact of AI in the music industry, going as far as to sue AI companies. But even UMG wants its data to help train generative AI as long as it can control who gets to use its copyright and build AI models.  UMG announced it will work with AI company Klay Vision to help train generative AI music models “ethically and fully respectful of copyright.” The model will help create commercially available AI-generated music that will also include protections for the likeness rights of human creators.   “Research is critical to building the foundations for AI music, but the tech is only an empty vessel when it doesn’t engage with the culture it is meant to serve,” said Ary Attie, founder and CEO of Klay, in a press release.   This is not the first time UMG has made a deal to offer its content for AI licensing, all the while allowing it to help shape how AI music models are built. UMG signed an agreement with YouTube to be part of YouTube’s parent Alphabet’s Music AI Incubator. It also worked with SoundLabs to help create voice models for artists.  Klay and UMG said the goal is to responsibly build AI music foundation models that the companies hope “will dramatically lessen the threat to human creators and stand the greatest opportunity to be transformational, creating significant new avenues for creativity and future monetization of copyrights.” The companies did not elaborate on what the music AI foundation model will do.  Famously litigious  The music industry is famously litigious, guarding its copyright and licenses tightly. After all, these are the same music labels that attempted to kill music downloads. The industry has been involved in the Congressional hearings on legally protecting copyright and the right to publicity of artists. UMG and the Recording Industry Association of America (RIAA), Sony Music and Warner Music Group’s Atlantic Records filed a copyright infringement claim against Suno and Udio. The labels alleged the two startups copied songs and that when prompted to generate similar-sounding songs, the platforms would return with the same songs with different lyrics instead. Both Udio and Suno denied the accusations.  Labels, once again including Universal, sued Anthropic, the company behind Claude,  for distributing lyrics without permission.  Preemptive participation As much as record labels want to protect their copyright, the partnerships the companies are forging point to a trend of industries adopting an “if you can’t beat ‘em, join ‘em” attitude towards AI.  Music labels see they can’t stem the creation of AI-generated songs and prevent AI models from training on publicly released music. Through these deals with AI startups, labels like UMG, which owns other record labels that host artists like Taylor Swift and Chappell Roan, can make (even more) money from their copyrights.  It also allows the companies to exert some control over who gets to use their data, something other industries have been pushing for as well. For example, media companies inked deals with companies like OpenAI, Perplexity, Google and now Meta, which just signed its first AI news partnership with Reuters.  Vickie Nauman, founder of music and tech consultancy Cross Border Works, said in an email to VentureBeat that when new technologies like AI “crash into the music sector, there is usually a burst of innovation alongside legal issues surrounding music rights.” “Major and smaller rightsholders all see that generative AI is here to stay and it is in everyone’s best interest to establish a sustainable legal market,” Nauman said. “The downside is these deals take negotiation and time, so it doesn’t happen immediately.” Music labels will undoubtedly continue to sue AI companies if they feel they are infringing on copyright, but recording agencies will also undoubtedly want to shape how AI music is created in the image that best suits them.  source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Microsoft unveiled an ambitious expansion of its artificial intelligence tools on Tuesday, introducing GitHub Copilot for Azure and a suite of developer-focused features that could fundamentally change how software is built in the AI era. The move represents Microsoft’s boldest attempt yet to dominate the rapidly evolving landscape of AI application development. At the heart of the announcement is a deceptively simple idea: eliminate the cognitive burden that developers face when switching between different tools and interfaces. It’s a problem that, according to Microsoft, costs developers an average of 23 minutes of productivity each time they context switch. “Developers today need to reach a heightened state of focus, because they’re creating a mental model about the application they’re trying to create. Having to interface with lots of different tools creates a huge amount of cognitive overload,” said Amanda Silver, CVP of Product for Microsoft’s Developer Division, in an interview with VentureBeat. The setup interface for OpenAI GPT-4 on GitHub guides developers through creating personal access tokens and integrating AI models into their workflows, reflecting Microsoft’s efforts to simplify AI implementation within coding environments. (Credit: Microsoft/GitHub) The rise of the AI engineer The timing of Microsoft’s announcement is particularly significant. As organizations rush to integrate AI capabilities into their applications, a new category of software developer is emerging — what industry insiders are calling the “AI engineer.” “If you think about the app workload from here on now, what developers are going to be doing, both in enterprises, commercial and even consumer, is going to be integrating intelligence into those applications,” explains Mario Rodriguez, Chief Product Officer at GitHub. “We’re seeing the rise of the AI engineer.” This shift represents more than just a new job title. It signals a fundamental change in how software is conceived, built, and deployed. Traditional software development follows a predictable pattern: code, build, debug, repeat. But AI development introduces new complexities, including model evaluation, prompt engineering, and managing the inherently probabilistic nature of AI outputs. A developer interacts with GitHub Copilot for Azure, using the AI-powered assistant to create and deploy an Azure Kubernetes Service (AKS) application, part of Microsoft’s initiative to streamline AI development within familiar coding environments like Visual Studio Code. (Credit: Microsoft) Breaking down the technical barriers Microsoft’s new tools aim to address these challenges head-on. GitHub Copilot for Azure acts as an AI-powered assistant that lives within popular coding environments like Visual Studio Code. It can help developers manage cloud resources, deploy applications, and even troubleshoot issues without leaving their primary workspace. The company is also introducing AI App Templates, which can be deployed “in as little as five minutes.” These templates support various AI frameworks and integrate with popular tools from vendors like Arize, LangChain, LlamaIndex, and Pinecone — a clear acknowledgment that AI development requires a diverse ecosystem of tools. For smaller teams and individual developers, these tools could level the playing field. “Experimenters and tinkerers can be very successful with all of these tools,” Silver noted. “When we think about the developer design point, it really is for creative developers exploring on their own.” The business implications The stakes are enormous. As enterprises race to integrate AI capabilities into their applications, the tools and platforms they choose today could lock them into specific ecosystems for years to come. Microsoft, with its ownership of GitHub and its Azure cloud platform, is uniquely positioned to capture this market. “We’re kind of at this stage right now where we’re starting to see Copilot go from single-threaded to multi-threaded,” Rodriguez explained. “We’re going from single model to multi-model… from single file editing to multi-file editing.” This evolution reflects a broader trend in the industry: the move toward more sophisticated, AI-powered development tools that can handle increasingly complex tasks. Microsoft’s announcement includes new capabilities for model evaluation and A/B testing at scale through GitHub Actions, allowing developers to automatically assess metrics like coherence and fluency as part of their deployment workflows. The road ahead While Microsoft’s new tools are impressive, they also raise important questions about the future of software development. As AI assistants become more capable, the line between human and machine contributions to code will blur. This could have profound implications for how we think about software authorship, liability, and intellectual property. Moreover, Microsoft’s integration of GitHub Copilot with Azure represents a significant advantage in the ongoing cloud wars with Amazon Web Services and Google Cloud. With 95% of Fortune 500 companies already using Azure, Microsoft’s enhanced developer tools could help it further consolidate its position in enterprise AI. The tools begin rolling out in preview this week as part of GitHub Universe, the company’s annual developer conference. Their success could determine not just Microsoft’s position in the AI race, but also how the next generation of software gets built. For developers, the message is clear: the future of software development is AI-first, and it’s arriving faster than many expected. As Silver puts it, these tools allow developers to “eliminate having to do the repetitive and the tedious and mundane and focus on the creative aspects of your job.” Whether this vision of AI-assisted development becomes the new normal will depend on how developers embrace these tools — and how Microsoft’s competitors respond to this latest evolution in the developer experience. source

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Moondream emerged from stealth mode today with $4.5 million in pre-seed funding and a radical proposition: when it comes to AI models, smaller is better. The startup, backed by Felicis Ventures, Microsoft’s M12 GitHub Fund, and Ascend, has built a vision-language model that operates with just 1.6 billion parameters yet rivals the performance of models four times its size. The company’s open-source model has already captured significant attention, logging over 2 million downloads and 5,100 GitHub stars. “What makes it special is that it is one of the tiniest models that is peculiar in its high accuracy, and it works just really well,” said Jay Allen, Moondream’s CEO and former AWS tech director. “It can run everywhere really easily and quickly. It can even run on iOS, on mobile phones.” Edge computing meets enterprise AI: How Moondream solves the cloud cost crisis The startup tackles a growing problem in enterprise AI adoption: the astronomical costs and privacy concerns of cloud computing. Moondream’s approach allows AI models to run locally on devices, from smartphones to industrial equipment. “As AI makes its way into more and more apps, I think we’re kind of torn between wanting all the benefits of the AI, but not necessarily wanting our entire lives broadcast to the cloud,” Allen told VentureBeat. “My preference is to do as much close to the edge so I have control over my own privacy.” Real-world applications: From retail inventory to factory floor intelligence Early adopters have found diverse applications for the technology. Retailers use it for automatic inventory management through mobile scanning. Transportation companies deploy it for vehicle inspections, while manufacturing facilities with air-gapped systems implement AI locally for quality control. The technical achievements stand out. Recent benchmarks show Moondream2 achieving 80.3% accuracy on VQAv2 and 64.3% on GQA — competitive with much larger models. The system’s energy efficiency impresses, with CTO Vik Korrapati noting “per token consumption is something like 0.6 joules per billion parameters.” David vs. Goliath: How a Small Team Takes On Tech Giants While major tech companies focus on massive models requiring substantial computing resources, Moondream targets practical implementation. “A lot of companies in this space are focused on AGI, and that ends up becoming a big distraction,” Korrapati said. “We’re laser focused on the perception problem and how we deliver cutting edge multimodal capabilities in the size and form factor that developers need.” The company now launches Moondream Cloud Service, designed to simplify development while maintaining flexibility for edge deployment. “What they want is the easiest path to start with a cloud-like offering so they can just play around with it,” Allen said. “But once they’ve done that, they don’t want to feel like they’re locked in.” This hybrid approach resonates with developers. The company has built a strong following in the open-source community, with Allen attributing this to their “hacker, open source ethos” and transparent development process. As for competition from tech giants, Allen remains confident in Moondream’s focused strategy. “For a lot of these large companies, this tends to be one of their 8,000 priorities,” he said. “There doesn’t seem to be a lot of companies that are as singularly focused as we are on providing a seamless developer experience around multimodal.” The company expects widespread enterprise adoption of vision language models within the next 12 months, though Korrapati cautions that “talking about timelines with AI is a dangerous game.” With the fresh funding, Moondream plans to expand its team, including hiring fullstack engineers at its Seattle headquarters. The company’s next challenge will be scaling its technology while maintaining the efficiency and accessibility that have defined its early success. source