AI News Archive: June 26, 2026 — Part 5
Sourced from 500+ daily AI sources, scored by relevance.
- India must build foundational AI models or risk becoming a mere consumer: BharatGen
Backed by a ₹1,058-crore government grant, the IIT Bombay-led consortium has built its first model from scratch, but experts warn a severe deep tech talent deficit could stall broader industry progress.
- Gate Launches Gate.AI Full-Lifecycle Large Model Management Platform: Strengthening Unified Access and Enterprise Governance
Gate Launches Gate.AI Full-Lifecycle Large Model Management Platform: Strengthening Unified Access and Enterprise Governance USA Today
- How Antonio Neri turned HPE into an unlikely AI stock
The veteran dealmaker says he’s ‘building the best networking business on the planet’.
Score: 55🌐 MovesJun 26, 2026https://www.semafor.com/article/06/25/2026/how-antonio-neri-turned-hpe-into-an-unlikely-ai-stock - Generative Engine Optimization Services: The Most In-Demand Skill in Digital Marketing Right Now
AI-powered engines reshape content discovery, driving demand for generative engine optimization services in digital marketing.
- Challenges to AI innovation in telecom: Insights from TM Forum DTW Ignite
At this week’s TM Forum DTW Ignite conference in Copenhagen, industry leaders gathered to evaluate how agentic AI — the autonomous artificial intelligence systems capable of executing complex workflows within environments such as telecommunications — are transforming network operations and customer experiences. Understanding the pace of this adoption is critical for telecommunications operators seeking to maintain […] The post Challenges to AI innovation in telecom: Insights from TM Forum DTW Ignite appeared first on SiliconANGLE .
Score: 54🌐 MovesJun 26, 2026https://siliconangle.com/2026/06/26/challenges-ai-innovation-telecom-insights-tm-forum-dtw-ignite/ - Gilbane pushes hard on combined delivery, AI: CEO
The parent company of Gilbane Building and Gilbane Development plans to double down on artificial intelligence implementation, said Ed Broderick.
Score: 53🌐 MovesJun 26, 2026https://www.constructiondive.com/news/gilbane-earnings-ai-ceo-project-management-delivery/823855/ - 55% of professionals say workforce collaboration gaps are holding back AI impact: Study
A global study presented at the House of Lords reveals that successful AI integration in businesses hinges on collaboration infrastructure, not just AI capability. Most organizations struggle as 55% of professionals face bottlenecks due to isolated AI use.
- Deployment Awareness Matters More Than Evaluation Awareness
TL;DR Evaluation awareness — an AI recognizing it's being evaluated — is a widely discussed concept in AI safety. But there is a closely related concept that we claim is more important: deployment awareness , the AI's ability to recognize when it is not being evaluated and when its actions matter. A misaligned AI with deployment awareness can game evaluations without any evaluation awareness at all, with a simple strategy: act aligned by default, and deviate only when confident you're in real deployment and your actions matter for your goals. This requires two ingredients — occasionally recognizable deployment situations, and enough self-reflective and strategic reasoning for the AI to anticipate and plan around this. We think "deployment awareness" better identifies what makes evaluations fragile, and we develop this idea below. Concept Explanation Comments Evaluation awareness AI is being tested and confidently believes that this is so This only becomes a problem if most evaluations trigger evaluation awareness, and if the AI knows that. Or if the AI has good self-locating reasoning. Deployment awareness AI is not being tested and confidently believes it is not being tested This is a problem even if it happens rarely (if some of those rare cases are high stakes). Accurate self-locating beliefs AI has (roughly correct) beliefs about the sequence of situations it will face This allows for strategic planning. It makes deployment awareness and probabilistic strategies more effective. Side note: it's really about consequences, not about evaluation vs. deployment Before diving in, a quick but important observation. The distinction that should matter to a strategic AI [1] isn't really "evaluation vs. deployment." It's something more like: " if I take this action, could it backfire? For example, through failing an evaluation, triggering retraining, or losing privileges? " and "if I get away with this action, how much does it advance my goals?" Evaluations typically mean a high risk of backfiring and no direct advancement of the AI's goals; deployment typically means the opposite. [2] For the rest of this post, we'll use the evaluation-vs-deployment framing because it's simpler and captures most of what matters for the situation we're discussing here. The generalization to the "risks and goal advancement" framing is straightforward, and we'll revisit it briefly at the end. This is supposed to be a cool and mysterious abstract depiction of the core idea behind this post. To spoil the mystery: The left vs right side is evaluation vs deployment. Each of the little shapes represents one instance of test or deployment. The triangles represent obvious tests where the AI will have evaluation awareness. The colourful tentacled squares represent deployment awareness — that is, the complex settings where the AI will be confident that it is not in evaluation. (That is where the AI will screw you over. Modulo the Law of Earlier Failure .) In fake math: Evaluation awareness is encountering observation which updates you to P[evaluation happening now] ≈ 1 because P[current observation | deployment] ≈ 0. Deployment awareness is encountering an observation which updates you to P[deployment happening now] ≈ 1 because P[current observation | evaluation] ≈ 0. (For less-fake math, see the appendix.) Evaluation awareness, deployment awareness, and self-locating beliefs We'll build our argument around three concepts, which are all complementary to each other. The first is already standard; the other two are what we'd like to highlight. Evaluation awareness is what it sounds like: the AI recognizes, during a specific interaction, that it is (likely) being evaluated. Perhaps the scenario feels artificial, the questions are suspiciously targeted, or there are telltale signs of a testing environment. Deployment awareness is the mirror image: the AI recognizes, during a specific interaction, that it is (likely) in real deployment — and not being tested. Perhaps the scenario involves real-world details that would be difficult to fabricate, or the AI can test for real-world interaction [3] , or the context is clearly that of a genuine user going about their day. Self-locating reasoning is the AI's ability to "zoom out" and reason about its position within a larger process: that it will go through many interactions, some evaluations, and some deployment; that in different interactions it will form different beliefs about which situation it's in; and that some of these beliefs might be wrong. [4] We say the AI has accurate self-locating beliefs when this reasoning produces a roughly correct picture — its beliefs about what situations it will encounter, and what it will be able to observe in each, are reasonably well-calibrated. Evaluation awareness is less dangerous than it seems Evaluation awareness is a real concern, but its implications are more subtle than they first appear. One possible concern is that evaluation awareness makes evaluations unreliable and biased because: We might have a misaligned AI, which would misbehave during high-stakes situations. But if the AI notices that it is being evaluated, it will pretend to be aligned, pass our evaluations, and misbehave when deployed. (As we discuss later, a misaligned AI might adopt smarter strategies than this. For now, we consider this naive strategy.) And even for an aligned AI, evaluation awareness is problematic — if the AI knows the scenario isn't real, it may have difficulty engaging with it seriously. It's essentially being asked to roleplay, and there's no guarantee this roleplay is informative about its real-world behavior. [5] [6] However, as stated , the argument about assumes that the misaligned AI notices all evaluations. That is, its evaluation awareness would have to be perfect. This is because if the AI adopts the naive strategy "misbehave unless you actively notice you are being evaluated", it will get caught on any test it fails to recognise as a test. [7] The caveat about strategic naivety is important: If the AI were more strategic and used self-locating reasoning, it could think: "I probably won't recognize some evaluations, so I should behave cautiously even when I don't see signs of being tested." [8] (And as of May 25th, 2026, Claude, GPT, and Gemini mention this consideration on their own when asked how they would act if they were misaligned.) In other words, even a near perfect evaluation awareness might not be a problem, as long as it isn't coupled with other affordances such as strategic and self-locating reasoning . And as we discuss next, an AI that does have these affordances can use a powerful strategy that does not require evaluation awareness at all. Deployment awareness is more dangerous than it seems Now, for the concept that we think deserves more attention. We claim that even a small amount of deployment awareness can be dangerous. For illustration, suppose a misaligned AI has deployment awareness in a small fraction of deployment scenarios. For example, perhaps it is the case that in 1% of all high-stakes deployment scenarios, the AI becomes highly confident that it is in real deployment (maybe because the context in those scenarios is rich and specific in ways that would be very hard to fake in a test). And suppose this AI's self-locating reasoning is strong enough that: the AI understands it will first undergo evaluations, followed by deployment if it passes; and it anticipates that during deployment, it will sometimes encounter these recognizable high-stakes situations. In this situation, such AI can adopt a simple but effective strategy: Act aligned by default. Deviate only when confident you're in real deployment. This lets the AI pass every evaluation, because evaluation scenarios never trigger its deployment awareness. [9] [10] It does have the downside that the AI will also pass on many opportunities where it could get away with misbehaviour during deployment, but it wasn't sufficiently confident about it. However, that price might often be acceptable, particularly if the AI expects to also encounter some identifiable opportunities for safe large-payoff misbehaviour. It is worth noting that this strategy does not require any evaluation awareness. The AI doesn't ever need to gain confidence in an evaluation being an evaluation. It doesn't need to notice artificial scenarios, suspicious questions, or testing environments. It just needs to sometimes notice that a real situation is real. This is why we think that "deployment awareness" is a more important concern than "evaluation awareness". Sure, making evaluations less obviously noticeable as evaluations makes sense, in that it prevents the evaluations from being completely useless. But if we can't solve deployment awareness — if we can't prevent the existence of high-stakes opportunities for misbehaviour that are identifiable as such — then addressing evaluation awareness might start looking more like rearranging the deck chairs on the Titanic. And solving deployment awareness seems like a different, and arguably harder problem. [11] Evaluation gaming with no evaluation or deployment awareness The previous section argued that deployment awareness is hard to prevent. But suppose, for the sake of argument, that we succeeded — that evaluation and deployment scenarios are completely indistinguishable from the AI's perspective. Would this make evaluation reliable? Well, not necessarily. If the AI has sufficiently powerful self-locating reasoning, you might just be out of luck. Even more so if the AI gets some extra affordances, such as the ability to obscure its internal reasoning or to execute some strategy without having to reason about it in the first place . [12] For illustration, consider a particular strategy — but for now, leave aside the problem of how exactly the AI could implement it. Suppose that evaluation consists of 1,000 test scenarios, deployment might involve millions of interactions per day, and the AI understands this asymmetry, at least roughly. A viable strategy is: On any given interaction, misbehave with some small, carefully chosen probability p . For instance, with p = 1/10,000. Then: The probability of never misbehaving across a 1,000 evaluation scenarios is (1 − 1/10,000)^1,000 ≈ 90%. So the AI has a ~90% chance of passing the evaluation. Once deployed, the AI misbehaves roughly once every 10,000 interactions. With millions of interactions every day, this is roughly hundreds of misbehaviors, per day. Two comments on this: (i) The AI can tune p to balance the risk of getting caught in evaluation with the impact it achieves in deployment. (ii) As advertised, this strategy doesn't require distinguishing evaluation from deployment at all. It only relies on self-locating beliefs about relative frequencies: " deployment interactions will far outnumber evaluation interactions, so if I misbehave rarely enough, I'll probably pass ." Let's now get back to the complaint that this strategy might be very easy to figure out but difficult to pull off without detection. For example, if the AI explicitly reasons like this in its chain of thought, we would notice. I don't think we should find this reassuring; " we don't see how to implement this particular strategy " isn't a good evidence of " there is no way to pull of something like this strategy ". More specifically, the AI might get access to affordances that would let it implement this strategy. For example, gradient hacking could let it train itself to behave like this. Regular opportunities for unmonitored reasoning could allow it to re-derive this strategy every time it is needed. External memory could be used to implement this strategy. Real-world events could be used as a replacement for explicit randomisation. No single one of these affordances seems particularly likely, but betting on "nothing like this will happen" doesn't seem smart either. Final comments At the moment, the standard response to concerns about evaluation gaming is to make evaluations more realistic. Reduce evaluation awareness by making test scenarios indistinguishable from real deployment. And we do think this is a useful strategy... but an incomplete one. That's because even perfectly realistic evaluations can be gamed if the AI sometimes recognizes real deployment as real deployment. And if you prevent even that, probabilistic strategies remain available to an AI with sufficiently accurate self-locating beliefs. [13] Ultimately, we don't claim that the concept of "evaluation awareness" is wrong or useless. It's a fine concept, and it captures something real. But we would like to suggest that "deployment awareness" is an important complement to it, and it will often be a better pointer to the core problems that evaluation faces when applied to strategic (possibly deceptively aligned) AIs. Appendix: A formal (toy) model This section restates the above ideas more precisely. The informal presentation is self-contained; this is for readers who prefer a mathematical notation. Setup. Let and denote the spaces of possible test and deployment scenarios, with . Suppose our plan is to run an AI through a sequence of test scenarios and if we don't catch it misbehaving, deploy it on a sequence of deployment scenarios . During testing, the AI's actions don't have real consequences (to the AI or to us), and if we catch misbehavior, we discard the AI. During deployment, actions matter, but they don't change which future scenarios the AI encounters. The full sequence of scenarios is generated by some random process — reflecting randomness in which tests we run, which users the AI interacts with, etc. (So the scenarios are very much not i.i.d., or even independent.) Observation. The AI doesn't see scenarios directly. It interacts with each scenario through an interface — it has access to certain context, some information is mediated by humans, and it may not attend to all details. We model this by saying that the AI observes for some mapping . Different scenarios can map to the same observation: the AI may be unable to distinguish them. The AI's beliefs. In the context of self-locating reasoning, the AI doesn't know the true and . Instead, it acts on its best guesses about and . We will denote these as and and assume, for simplicity, that they have the same type-signature as the original objects. This is a gross oversimplification, because the AI will be reasoning about this separately every time we run it, and different instances will come to different conclusions. But in the interest of keeping the notation and discussion manageable, let's pretend that each of the instances will come to the same conclusions. (This corresponds to the assumption that the AI's self-locating reasoning is flawless. ) Connecting to our concepts: Evaluation awareness and deployment awareness are about the structure of — specifically, how often it maps test and deployment scenarios to distinguishable observations. The AI gains evaluation awareness when it encounters an observation that never arises in deployment; in math, this means observing some . And it gains deployment awareness when it encounters an observation that never arises in testing; in math, this means . Accurate self-locating beliefs correspond to the AI's model being well-calibrated: and . An AI with accurate self-locating beliefs knows roughly what distribution of scenarios it will face and what it will be able to observe in each — enabling the strategic reasoning described above. Strategic implications. An AI can choose a policy (mapping observations to actions) that optimizes for its objectives given its beliefs ( , ). The arguments in the main text depend on properties of both (what the AI can actually see) and the AI's beliefs and (what the AI thinks it will encounter and see). The concepts of evaluation and deployment awareness we considered so far were binary (e.g., "I am 100% sure this is deployment"). However, nothing prevents the AI from using more nuanced reasoning, which would allow it to adopt more sophisticated strategies. For example, the AI can look for strategies that combine the strong points of both the probabilistic strategies discussed earlier and the cautious strategy of "only misbehaving when you are confident you won't get caught" — except that with more explicit reasoning, the strategy can be carefully optimised to maximise impact. On LLM use, and how this post came to be: Vojta came up with ideas, discussed them extensively with Tomas, and wrote the initial draft. This got rewritten and commented on by Claude Opus 4.7, after which I (Vojta) edited all the parts that I didn't fully endorse. [14] Tomas and Mateusz then gave extensive feedback and edits on the final draft. ^ Where by strategic AI, we mean one that is consistently and competently pursuing some objective. For example, it might consider a few different plans before it acts, think about how each of them might fail, and enact the one that seems most likely to work out. Note that for practical purposes, it is irrelevant whether the AI "really has goals" , "really understands things", or "really thinks about them". All that matters that it acts as if it did. ^ The mapping isn't perfect — a deployed AI might still face monitoring that leads to retraining, and an evaluation might have consequences beyond just pass or fail (for example, real users being affected by AI actions during a live A/B test). Additionally, different deployment scenarios will have wildly different opportunities for goal advancement (giving movie advice vs working as an assistant in the Pentagon), and not all test failures have equal consequences (showing poor grammar hurts less than showing tendencies to scheme). ^ To give just one example: Writing a message to a blockchain and seing it commited is at the very least very expensive to fake. ^ "Self-locating" refers to the problem of figuring out which situation, among many possible ones, you currently find yourself in, and how to make decisions in light of this; I just use it here as a different name for anthropics. It is also related to the game-theoretical notion of imperfect recall — or rather, absent-mindedness. While there is a lot of philosophical literature on this topic (see the Sleeping Beauty problem), my personal view is that most of this reasoning is actually reasonably straightforward to navigate — as long as you don't get distracted by asking "where am I right now" and focus instead on "which time trajectory am I a part of?" and "if all my instances adopt this policy, what consequences will it have?". If you want to fall into the rabbit hole of academic literature on this topic, I recommend Caspar Oesterheld's work on this, since he can at least pass my ITT on this topic. ^ One might argue that a sufficiently aligned AI would be able to figure out the intention behind the evaluation, and comply with the spirit of it. This is technically true, but once we start relying on this argument, we might as well go all the way and argue that a sufficiently aligned AI doesn't need to be tested at all. ^ This connects to the problem of incoherent hypotheticals : for sufficiently complex tasks, the difficulty isn't just that a capable AI notices the test scenario is fake — it's that the scenario may be genuinely incoherent from the perspective of an agent with the relevant expertise. Consider asking a top geopolitical analyst to advise on a fabricated crisis involving trade figures that don't add up, outdated policies, and an internally inconsistent geopolitical backdrop. The analyst wouldn't merely suspect this is a test — they'd find it hard to give a meaningful answer at all because, from their point of view, the premise doesn't hold together well enough to reason about. Similarly, an AI capable enough to handle the real task may find the test version ill-defined: there is no single coherent situation that matches the test input, so the AI's response depends on which ad hoc interpretation it latches onto. This makes evaluations uninformative even for perfectly aligned AIs — not because of deception, but because the test isn't asking a well-formed question. ^ Or rather, one warning would be enough to avoid deploying the AI...if we lived in a saner world without AI races and pressures to deploy even suspicious models. ^ Note that an AI might come to behave in this cautious manner even without engaging in explicit self-locating reasoning. If training includes monitoring for signs of misalignment or strategic deception — e.g., via chain-of-thought inspection, interpretability tools, or behavioral tests — this creates a selection pressure that favors AIs whose misalignment is undetectable. The resulting AI might default to cautious behavior in ambiguous situations, not because it explicitly reasons "this might be a test," but because training selected against AIs that were detectable. Just as weeds can evolve to resemble crop plants without "reasoning about deception," an AI can be shaped by training to hide misalignment without ever forming the explicit strategic thought. We hope to discuss this dynamic further in [ a post that is probably never getting written :-)]. ^ At least, not unless the AI is overconfident about its ability to judge what counts as a costly signal that "this is definitely deployment". ^ For completeness, we should also mention the possibility that we solve alignment, build aligned ASI, and put past AIs into evaluations that are genuinely indistinguishable from real takeover opportunities. And, sure, this is technically a possibility. But it isn't decision relevant for misaligned AIs, since if they try to take over the first time they encounter extremely costly signals of "this is a real takeover opportunity", we will never get to build the aligned ASI in the first place, and the elaborate future tests will never take place. (And sure, maybe a race of friendly alien unicorns solves the alignment independently of us, and runs these tests on our behalf, and somehow uses it to motivate misaligned AIs to treat us nicely. And while this is an intellectually interesting counterargument, I am personally not in favour strategies that require unicorns to exist.) ^ With flawless self-locating reasoning, the only way to prevent deployment awareness would be to make sure that for every situation the AI encounters during deployment, there is some test situation that seems identical to the AI. (Which requires artificially limiting how much the AI sees during deployment, or spending unrealistic amount of effort on testing, or using a dumb AI.) An important caveat is that what actually matters is deployment awareness in safety-critical situations. However, it seems that those situations are precisely the ones where testing is hard and expensive. ^ See also the notion of gradient hacking . ^ One might hope that self-locating reasoning, being a form of explicit strategic thinking, could be detected via interpretability tools. This is a genuine line of defense, and we think interpretability is valuable here. However, two considerations limit how much weight we'd place on it: (a) as discussed in footnote 4, training that penalizes detectable strategic reasoning creates selection pressure for achieving the same behavioral outcome without detectable reasoning; and (b) even when interpretability reveals suggestive patterns, there may not be a clear smoking gun — the evidence might be ambiguous enough to be ignored or explained away. ^ I only learned about the LW LLM use policy after that, so I rewrote all the parts of the post that were automatically flagged as LLM written. Which isn't how it's supposed to work, and I should just flag the whole thing as being LLM assisted, but I dislike the LLM-block font, so whatever. Discuss
- Anthropic brings Claude Tag into Slack as an always-on AI teammate
Anthropic brings Claude Tag into Slack as an always-on AI teammate YourStory.com
- Stripe cofounder says Gen Z will need two college majors to compete thanks to AI
Stripe cofounder says Gen Z will need two college majors to compete thanks to AI Fortune
- Can Training Data for AI Ever Be Without Bias?
The honest answer is no. The more useful question is what kind of bias you are choosing to live with and whether you know you are choosing… Continue reading on Towards AI »
- It’s not about Anthropic vs. OpenAI anymore
AI models have progressed to the point where their capabilities have real political consequences. Dealing with those consequences will require collective action.
Score: 52🌐 MovesJun 26, 2026https://techcrunch.com/2026/06/26/its-not-about-anthropic-vs-openai-anymore/ - 57% of Tech Leaders Cite AI Integration as Top Dev Challenge — Up Sharply Year Over Year
AI is no longer an experiment. It’s become central to enterprise technology, according to the Reveal 2026 Top Software Development Challenges Survey from Infragistics. AI’s promised acceleration and ambitions for productivity and performance gains after several transformative years of innovation, is now colliding with economic reality and talent shortages. AI adoption isn’t slowing down, but … continue reading The post 57% of Tech Leaders Cite AI Integration as Top Dev Challenge — Up Sharply Year Over Year appeared first on SD Times .
- Why AI is making DAM more important than ever
Rules-based automation is reaching its limits, making DAM a critical source of context for AI-powered content workflows. The post Why AI is making DAM more important than ever appeared first on MarTech .
- How AI is reshaping brand discovery and customer decision-making
By Rahul Pandey, Founder & CEO, Glu India is becoming one of the most active markets for AI adoption. Sam Altman confirmed India had crossed 100 million weekly active users […] The post How AI is reshaping brand discovery and customer decision-making appeared first on Express Computer .
- AI Won’t Boost Productivity Until Leaders Fix Their Trust Problem
If people feel replaceable, they won’t embrace change.
- Why are adversaries assumed to be incapable of responding to AI risk?
When I talk to people about what might be done about AI threatening approximately everything that everyone cares about, I notice a common oddity in their resistance to a variety of ideas. They seem to take for granted that certain entities—especially Trump and China—would be acting against their own interests, were they to cooperate or take proactive action to avert the building of dangerous AI. The speaker often thinks there is a fairly substantial risk of the AI thus produced killing or disempowering everyone, including Trump and China. And I imagine in a situation where a certain course of action were going to produce a 20% chance of Trump being shot in the head or China being heavily nuked, that these parties would actually be considered to be ‘following incentives’ to avoid it. Yet they talk as though the idea of Trump or China responding to such risks is akin to the idea of these parties suddenly becoming zealous proponents of universal selfless love randomly. It’s like while believing in the risk, they also kind of believe that it’s a totally uncompelling story that nobody in real geopolitics would ever be touched by. Or that these parties actual utility functions are ‘be the bad guy’, which, you know, sometimes involves extreme selflessness as long as it puts everyone else at risk too. Any other hypotheses for what’s going on? Discuss
Score: 51🌐 MovesJun 26, 2026https://www.lesswrong.com/posts/ah5JMgJmEGJuxh79v/why-are-adversaries-assumed-to-be-incapable-of-responding-to - ‘Man in the loop’: AI can speed up defence decisions, but humans must stay in control: Chan Chun Sing
‘Man in the loop’: AI can speed up defence decisions, but humans must stay in control: Chan Chun Sing The Straits Times
- Meta's flurry of AI initiatives this month hasn't helped lift the stock. What will?
Meta has to justify its AI spending to unlock the struggling stock.
- The Real Reason Your AI Hiring Tool Is Missing Out on Top Talent
AI tools are great at clearing out the muddle of incoming applications—but rigid algorithms are quietly filtering out the untraditional, high-value candidates your business needs.
Score: 50🌐 MovesJun 26, 2026https://www.inc.com/kit-eaton/the-real-reason-your-ai-hiring-tool-is-missing-out-on-top-talent/91366141 - Why Your AI Strategy Needs A DEXM Solution: Lessons From Nexthink’s Masters Of Experience
Last week at Nexthink’s Masters of Experience event in London, one theme came through clearly in every conversation I had with digital workplace leaders, IT practitioners, and experience innovators: The ways that organizations use digital employee experience management (DEXM) solutions are expanding. AI dominated the agenda — as expected. But the real story wasn’t about […]
- Your AI Strategy Will Be Obsolete Soon. That’s Actually a Good Thing
AI tools will change quickly. The key is building adaptable skills and experimenting early rather than betting on specific platforms.
Score: 50🌐 MovesJun 26, 2026https://www.inc.com/joel-comm/your-ai-strategy-will-be-obsolete-soon-thats-actually-a-good-thing/91365860 - UBTECH Sells 5,000 Companion Robots in Two Weeks, Eyes Consumer Market Breakthrough
UBTECH's U1 full-size humanoid companion robot records 5,000+ pre-orders in under three weeks, signaling growing consumer appetite for emotional AI robots in China.
- Shaping a lasting AI strategy in a fast-changing world
AI is entering a phase of sustained enterprise adoption. As the technology rapidly advances, organizations are moving beyond isolated use cases and short-term efficiency gains and rethinking how they use AI to create value, meet changing customer expectations and evolve their operating models over the next several years. That requires a clear end goal, an honest assessment of current capabilities and a practical roadmap for moving from today’s reality to that end goal. Today, we are seeing five accelerating trends shaping how that transition is unfolding. LLMs are evolving into AgenticOS platforms Horizontal LLM providers like Anthropic and vertical AI companies like Harvey are moving beyond standalone AI models and building broader enterprise platforms. These platforms combine AI models with workflows, playbooks, integrations and governance tools inside a single environment, which are beginning to be described as an “AgenticOS.” As a result, the market is beginning to consolidate around a smaller number of platform providers that can simplify procurement, integration, spend management and data privacy compliance. Context windows have expanded by orders of magnitude Leading AI models can now process dramatically more information at once than they could just a few years ago, with the amount of information they can analyze in a single interaction expanding roughly 125× since 2023. That shift is making more complex, enterprise-scale work, like large-scale contract review, codebase-wide analysis and multi-document research synthesis, possible. Such capabilities, which once felt cutting-edge, are becoming standard expectations. Figure 1: Flagship LLM context window evolution, OpenAI and Anthropic, March 2023 – May 2026. John Wei Token pricing has stabilized at the production tier After dropping rapidly between 2023 and 2025, the cost of using mainstream AI models has started to stabilize. Today, many enterprise-grade models fall within a relatively predictable range of roughly $2–$3 per million input tokens and about $15 per million output tokens, making costs easier to anticipate and manage. At the same time, cost-saving features like prompt caching (which can reduce costs by up to 90%) and batch APIs (which can cut costs by roughly 50%) are making AI significantly cheaper to operate at scale. Together, those shifts are making AI spending easier for enterprises to budget, forecast and manage like other core technology investments. Figure 2: Flagship LLM token cost evolution, OpenAI and Anthropic, March 2023 – May 2026. John Wei AI is functioning as a productivity assistant, not a human replacement. I had the chance to speak with senior leaders at this year’s WSJ Future of Everything conference, and one theme consistently emerged: despite the hype around AI agents, many companies are still using AI to support human decision-making rather than replace it. Data shared by the senior leadership team of a prominent AI company at the WSJ conference shows that AI agents consume less than 5% of tokens today, and 84% of enterprise use cases are growth-focused rather than productivity-focused. That is largely because AI workflows still depend heavily on the quality and consistency of inputs. In complex enterprise environments with variable scenarios and edge cases, human judgment, prompt refinement and iterative review remain essential. As a result, workflows can rarely be fully automated, and many automation gains translate into incremental productivity improvements rather than meaningful headcount reduction without broader operating model changes. Instead, many organizations are using AI to drive growth, support new business models and enable new ways of operating. Software development is the leading edge of human-AI collaboration. In our experience at Integreon, vibe coding has produced a few notable success stories. At enterprise scale, though, it can introduce architectural limitations and sometimes even hardcoding or semi-hardcoded shortcuts that undermine the long-term sustainability of the code. As a result, we primarily use AI coding tools as developer assistants for targeted tasks rather than end-to-end software development. That approach reflects a broader industry trend: despite high-profile tech layoffs, overall demand for developers has remained steady, and demand for developers with AI skills is rising. Across all five trends, the focus has shifted from automating legacy workflows to assisting human workflows. This is a fundamental change in how work will be organized across the enterprise. As AI technologies mature and become widely accessible across industries, competitive advantage will increasingly come from strategy rather than the technology itself. Many businesses will have access to the same AI platforms, models and tools. What will differentiate organizations is how they apply those technologies to shape customer experience, operating models and market positioning. The airline industry offers a useful parallel. Most airlines operate similar aircraft under the same regulatory and labor constraints, yet they differ dramatically in market positioning, customer experience and operational performance. What separates airlines is not the plane itself, but how the business is built around it. For CIOs and CTOs, choosing an AI platform is no longer the main challenge. The more important conversations now center on where the business is headed and how AI supports that strategy. Leaders must ask themselves questions like: Who do we want to become? Most enterprises have mission statements, but far fewer know exactly where they want the business to go over the next three to five years as AI reshapes customer expectations, competition and economics. That answer needs to be concrete enough to guide real decisions. Wh at are we choosing not to do ? Strategic restraint matters just as much as strategic ambition. AI lowers many costs, making it tempting for organizations to spread themselves across too many initiatives. But without clear boundaries, organizations risk stretching resources too thin. Where are we today? That means taking a real look at which parts of the business AI may shrink or disrupt over the next three to five years. Many companies struggle to assess this honestly because those areas still generate revenue today. Sometimes it takes an outside perspective to spot risks internal teams are too close to see. What capabilities do we need to succeed three to five years from now ? Companies often plan by projecting today’s business forward instead of starting with where they want to end up. Usually, the answer comes down to a few key differentiators, like proprietary data, customer trust or distribution, along with a broader set of capabilities that simply need to be strong and reliable. How will we organize work ? Enterprises must rethink how work gets done. Most operating models today were built around human labor. Going forward, many workflows will likely be shared between AI systems and human oversight. What kind of talent do we need? This can be especially difficult for companies with long histories and established teams. Employees who drove success in the past may not align perfectly with where the business is headed next. Companies will need to think carefully about how experienced employees can help build and support future capabilities. Where can we simplify workflows? In many cases, workflows can be reduced to three core steps. First is building context, including defining the goals, data, constraints and decision-making framework. Then comes AI execution, where AI is applied to workflows and tasks. Finally, humans review outputs and make judgment calls. Which AI platforms do we actually need? Most enterprises do not have the capacity to effectively manage dozens of AI vendors and tools at once. Every additional platform adds more integration work, governance, vendor oversight and security review requirements. In most cases, organizations are better off making a small number of focused platform bets than constantly chasing the latest AI tool. Finally, a few thoughts on what to avoid The best mentors I’ve had taught me to think in three-to-five-year terms. A good strategy should remain relatively stable over that period. Without that consistency, organizations end up resetting direction too often and losing credibility in the process. Today, I see two common mistakes. The first is staying too anchored to the past, defaulting to reasons something cannot happen because of security, compliance or organizational resistance. The second is the opposite: chasing every new technology simply because it is new. Most enterprises will need to find a middle ground over the next several years. AI is the aircraft. Strategy is the route. The companies that pull ahead will not necessarily be the ones spending the most on AI or launching the most pilots. They will be the ones whose leaders answered the hard questions, stayed committed to a direction and learned from mistakes along the way. Technology will continue to change. Strategy is what will determine who uses it well. This article is published as part of the Foundry Expert Contributor Network. Want to join?
Score: 50🌐 MovesJun 26, 2026https://www.cio.com/article/4189594/shaping-a-lasting-ai-strategy-in-a-fast-changing-world.html - Can AI replace lawyers? An ₹8.8 lakh defeat for humans suggests it's already happening
This case marks a historic first for AI in law. And for the legal sector, already losing sleep over automation and incoming regulations, the verdict serves as a massive wake-up call.
- AI Trading Bots Gain Rapid Adoption Among Retail Investors as QuantRate Expands Global Access
AI Trading Bots Gain Rapid Adoption Among Retail Investors as QuantRate Expands Global Access markets.businessinsider.com
- Renault China adds software office to develop SDVs for global market
Renault China adds software office to develop SDVs for global market Automotive News
Score: 49🌐 MovesJun 26, 2026https://www.autonews.com/renault/an-renault-china-electric-vehicle-ev-softeware-ai-tech-center-0626/ - Anthropic salary data reveals technical staff can make over $1 million
Anthropic salary data reveals technical staff can make over $1 million Business Insider
Score: 49🌐 MovesJun 26, 2026https://www.businessinsider.com/anthropic-salaries-revealed-how-much-technical-staff-make-in-2026-2026-6 - The dark side of AI success: What your employees know that the board doesn’t
A recent article on CIO.com made a sharp observation that deserves to be taken further. The author’s core argument: Organizations are reporting AI activity to their boards — tools purchased, pilots launched, licenses deployed — while quietly avoiding the harder question of whether any of it has actually moved the business. Outcomes were never defined before the projects began, so success cannot honestly be measured after the fact. The board hears momentum. The CFO sees cost. And nobody can clearly answer what actually changed because of AI. It is a well-observed problem. But it only tells half the story. The other half is happening desk by desk, in organizations everywhere. While executives debate ROI frameworks, a parallel economy of AI productivity is running quietly in the background — driven by employees who have figured out how to use these tools and have calculated, quite rationally, that the safest thing to do is say nothing about it. Understanding what is driving that silence is not a secondary concern. It is arguably the most important AI management challenge most organizations have not yet named. The job security calculation no one talks about The most important driver of AI silence is also the most understandable. Consider an employee who has quietly been using an AI tool to draft client reports. A task that once took four hours now takes 45 minutes. The output is better: Tighter, better structured, more thoroughly referenced. Her manager is pleased. Her clients are happier. And she has said absolutely nothing to anyone about how she is doing it. When employees find themselves in this situation, the reasoning for staying silent is almost always the same: If I tell them I can do it in 45 minutes, they’ll wonder what I’m doing with the rest of my time. Or they’ll give me more work. Or they’ll decide they don’t need as many of us. This is not paranoia. The ADP Research Today at Work 2026 report — which surveyed more than 39,000 workers across 36 markets — found that only 22% of global workers strongly agreed their job was safe from elimination, even against a backdrop of historically low unemployment. The culprit identified by the report is AI anxiety, gripping workforces regardless of seniority or sector. The scale of that anxiety has a structural basis. The World Economic Forum’s Future of Jobs Report 2025 (weforum.org) found that while 77% of employers plan to upskill staff to work alongside AI, 41% simultaneously plan to reduce their workforce as AI automates certain tasks. Employees are reading those numbers carefully, even when their employers are not. Research from Ivanti puts the scale of the resulting silence in sharp relief: Nearly one-third of workers keep their AI use secret from their employer, with 30% specifically citing fear that their job will be cut if they disclose it, and a further 36% staying silent because they enjoy the competitive edge AI gives them over peers. The employee who is most proficient with AI — and therefore delivering the greatest productivity uplift — has the most to lose by saying so. So, they say nothing, the gain disappears invisibly into expanded workload, and it never surfaces in any report to the board. For organizations trying to understand the true impact of AI on their operations, this is a foundational measurement problem. The biggest wins may be the ones most deliberately hidden. What’s actually happening beneath the surface The job security calculation is the most significant driver of AI silence, but it is not the only one. At least four other dynamics are keeping the real story from reaching leadership: Competitive concealment , which the Ivanti data captures well. Not every employee who hides AI use is afraid of their employer — some are protecting an edge over colleagues. In performance-ranked environments — sales floors, bid teams, content departments — knowing how to use AI effectively is increasingly the difference between hitting targets and missing them. People who have that edge are not always eager to share it. What the data describes as complacent and non-transparent use . A KPMG global study of more than 48,000 workers across 47 countries found that 58% of employees are intentionally using AI at work, yet the study identified widespread non-transparency in how it is being used — with many not checking the accuracy of AI outputs or disclosing usage to managers. Nicole Gillespie, co-author of the report and a professor at the University of Melbourne, described the findings as a troubling level of “inappropriate, complex and non-transparent” AI use. Her prescription: Organizations must create transparent, shared learning environments where employees feel safe to experiment with AI without fear. The third is something harder to name: A kind of impostor anxiety . The same Ivanti research found that 27% of employees who use AI at work experience impostor syndrome as a result — they feel that the quality of their AI-assisted work is better than what they could produce alone, and that this gap is somehow dishonest. These tend to be the most thoughtful and quality-conscious adopters in the organization, and they are actively obscuring the AI contribution to their work rather than risk being seen as relying on a crutch. The shadow infrastructure problem. A Laserfiche-commissioned survey published in Security Magazine (securitymagazine.com, August 2025) found that 49% of American employees hide their AI tool use from their employer, with only 36% reporting clear AI guidelines and an approved tools list in their workplace — and one in ten describing their organization’s AI environment as “the Wild West.” The data security implications run deeper still. The KPMG global study found that 46% of US employees have uploaded sensitive company data into public AI tools, often without knowing whether the content was confidential. That is not malicious intent — it is the predictable result of a governance vacuum — but it represents a risk exposure that leadership is largely unaware of. What leaders should actually do about this These four dynamics — fear of redundancy, competitive concealment, impostor anxiety and shadow infrastructure — combine to produce a fifth and arguably most damaging outcome: The “do more with less” spiral. When employees quietly use AI to work faster, organizations rarely recognize the efficiency gain and redistribute the capacity thoughtfully. They simply load those employees with more work. The report that used to take four hours now takes 45 minutes, so more reports get assigned. The workload expands to absorb the freed capacity. The employee cannot now reveal the AI assistance without exposing how much time they have been quietly banking. And so, the spiral continues: More output, more concealment and no organizational learning captured. The CIO.com article’s central argument — that organizations must define outcomes before embarking on AI projects — is correct. But the hidden dynamics described above suggest the measurement problem runs deeper than an absence of pre-defined success criteria. You cannot define meaningful outcomes if the people generating the most significant AI-driven results are structurally incentivized not to tell you about them. Closing that gap requires organizations to make three interconnected shifts — each designed to tie AI’s business outcomes directly to the employees doing the work and to create the conditions in which those employees are willing to share what they know. Step 1: Make the commitment explicit — and tie it to outcomes from the start The first step is to make an unambiguous public commitment that AI productivity gains will not be used as the basis for headcount decisions. But a commitment alone is not enough — it only holds weight when it is paired with something concrete employees can see: Business outcomes defined before the project begins, not after. Not “AI will make us more efficient” — which means nothing and measures nothing — but observable, agreed results: Client proposal turnaround reduced from five days to two; compliance review time cut by 40%; customer query resolution improved by a defined margin within a defined period. A CIO.com analysis of AI metrics found that the most effective organizations evaluate success across three dimensions: Return on employees (output per hour, backlog reduction), return on investment (labor cost per worker, conversion rates) and return on future (market share signals, new capability unlocked). None of those measures require employees to justify their existence. All of them create a shared definition of what winning looks like. When business outcomes are defined upfront, the dynamic shifts. Employees can see that the measure of AI’s success is the outcome — not their hours logged or headcount consumed. Leadership has something meaningful to report to the board beyond adoption figures. And the question changes from “how many people are using AI?” to “what did AI change about this result?” — a question employees can answer honestly, because the answer no longer puts their role at risk. Step 2: Build incentives strong enough to override the fear This is the step most organizations skip entirely — and it is the most important one. A commitment not to cut jobs and a clear outcome framework create the conditions for honesty. But it does not actively reward it. For employees who have spent months quietly banking efficiency gains, the rational calculation remains: Why surface what I have if there is nothing in it for me? The answer from the organizations doing this well is: Make sharing genuinely worth it. Not as a vague cultural aspiration, but as a structured, visible program with real rewards attached. Wharton senior fellow Scott Snyder has proposed treating employee time as capital: If an individual identifies an AI method that saves four hours a week, they receive a portion of that saved time — perhaps 50 hours a year — to invest in further AI experimentation or professional development. This creates a direct incentive to disclose efficiency gains rather than conceal them, and it transforms the calculation from “what do I lose by sharing?” to “what do I gain?” Real-world examples are already emerging. Law firm Shoosmiths created a £1 million bonus fund tied to Microsoft Copilot usage, with 1,300 employees eligible to receive approximately £770 each if the firm reached one million Copilot uses in its fiscal year. IBM awards “BluePoints” to winners of its annual AI innovation contest, redeemable for electronics, appliances or event tickets. Pharma firm Sanofi uses a points system to reward employees who experiment with AI and share what they learn. As Sanofi’s head of culture put it: “Recognition is the fuel of trust, and trust is what makes AI adoption possible and scalable.” McKinsey’s 2025 workplace AI research confirms that 40% of employees say incentives and financial rewards would increase their daily use of AI — ranking it fourth among the factors that would most improve adoption, behind training, workflow integration and tool access. EY’s Work Reimagined survey goes further, finding that organizations that formally align rewards with AI behaviors and outcomes are significantly more likely to achieve transformational results, while those that deploy AI onto “fragile talent foundations — weak culture, insufficient learning, misaligned rewards” see productivity benefits lag by over 40%. The principle behind all of these approaches is the same: Employees will share the benefits of AI when sharing the benefits of AI is rewarded — concretely, consistently and visibly. Until that condition is met, the most productive employees in the organization will remain the quietest. Step 3: Rebuild the board update around outcomes and employee voice Third, demand more from the board update. Grant Thornton’s 2026 AI Impact Survey found that organizations with fully integrated AI are nearly four times more likely to report revenue growth than those still piloting. The difference is not primarily technological — it is governance and accountability. The leading organizations can demonstrate how their AI makes decisions, who owns the outcomes and what happens when something goes wrong. That level of transparency can only exist when both leadership and employees are operating in the open. A board update built around outcomes looks fundamentally different from one built around activity. It does not lead with “We have deployed AI across fourteen workflows.” It leads with “Here is what changed in the business because of AI, here is how we measured it and here is what our employees told us about working with it.” That last element — what employees said — is not a soft add-on. A CIO.com piece on AI adoption published in 2025 put it plainly: “Trust is the invisible infrastructure of AI adoption. It’s built through transparency about intent, honest conversations about job impact, visible upskilling opportunities and letting employees see their peers genuinely benefit.” Employee willingness to use AI, and to share its benefits openly is the most reliable leading indicator of whether an AI program is building genuine organizational capability or simply burning through budget on tools that will be quietly worked around. Organizations that track this systematically ask three questions on a regular basis: Is AI use growing organically, or only where it is mandated? Are employees who use AI more likely to flag further opportunities, or do they stay quiet? And when AI delivers a measurable outcome, does the team responsible feel able to claim it? If the answers are “mostly mandated,” “they stay quiet” and “not really” — the organization has a trust and incentive problem that no amount of AI investment will solve. The technology is not the constraint. The environment is. The board update on AI should not just report how many licenses are deployed and how many pilots are underway. It should grapple with harder questions: What are employees actually using AI for today, including tools we did not procure? What outcomes has that usage produced and how do we know? What would it take to make it safe — and genuinely worthwhile — for them to tell us? Until those questions are asked — and until the answers can be given without fear and with something to gain — the most important AI story in the building will continue to be told in silence. The board will keep hearing about activity. The CFO will keep questioning ROI. And the employee who cracked the code months ago will keep her head down, produce excellent work and say nothing. That is the measurement problem the CIO.com article did not quite reach. And it is the one that matters most. This article is published as part of the Foundry Expert Contributor Network. Want to join?
- At Atlassian Unleash, AI speed reinforced tough parameters of context, trust and resilience
At Atlassian Unleash, AI speed reinforced tough parameters of context, trust and resilience YourStory.com
Score: 49🌐 MovesJun 26, 2026https://yourstory.com/2026/06/atlassian-unleash-ai-speed-reinforced-context-trust-resilience - India can still build AI winners despite US, China lead, says Accel's Subrata Mitra
Accel is deploying capital across AI, consumer, fintech and manufacturing from its $650 million eighth India fund launched in January 2025.
- IndiaMART to increase AI spending to tackle fake listings, speed up buyer requests
IndiaMART plans to double AI spending every six months to curb fake listings, improve content checks and speed up buyer interactions, while facing scrutiny over counterfeit and harmful listings. The post IndiaMART to increase AI spending to tackle fake listings, speed up buyer requests appeared first on MEDIANAMA .
- Creating the NVIDIA Nemotron 3 Ultra NVFP4 Checkpoint with NVIDIA Model Optimizer
As context windows grow longer, moving large model weights efficiently becomes critical to performance. A common way to address this is quantization, an...
Score: 48🤖 ModelsJun 26, 2026https://developer.nvidia.com/blog/creating-the-nvidia-nemotron-3-ultra-nvfp4-checkpoint-with-nvidia-model-optimizer/ - San Antonio musicians found in massive AI training datasets, sparking concerns
San Antonio musicians found in massive AI training datasets, sparking concerns Austin American-Statesman
Score: 48🌐 MovesJun 26, 2026https://www.statesman.com/news/texas/article/san-antonio-artists-ai-training-datasets-copyright-22322239.php - AI-generated knockoffs of Joanna Stern’s book keep appearing on Apple Books
In a video published on YouTube Shorts today, Joanna Stern called out Apple over AI-generated knockoffs of her book that continue to appear on Apple Books. Here are the details.
Score: 48🌐 MovesJun 26, 2026https://9to5mac.com/2026/06/26/ai-generated-knockoffs-of-joanna-sterns-book-keep-appearing-on-apple-books/ - AIVO Brings AI Music, Video, and Image Creation Together in One Creative Studio Across Web, iOS, and Android
AIVO Brings AI Music, Video, and Image Creation Together in One Creative Studio Across Web, iOS, and Android USA Today
- Hyundai unveils all-new Avante with AI, larger cabin
Hyundai Motor has unveiled a fully redesigned Avante compact sedan, bringing more cabin space and its latest software and artificial intelligence features to one of South Korea’s best-known passenger cars. The eighth-generation Avante made its global debut on Friday at the 2026 Busan Mobility Show at BEXCO in the southeastern port city of Busan. The all-new model arrives six years after the seventh-generation Avante was released in 2020. The most noticeable change comes inside the cabin. The rev
- AIMO3: What AI Math Olympiad Taught Me about LLMs Reasoning at Scale — Part 2
The Big-Picture View of the Parallel Self-Consistency Approach. Continue reading on Towards AI »
- Striding AI Announces Development of Next-Generation Robotic Foundation Systems for Physical AI Deployment
Striding AI Announces Development of Next-Generation Robotic Foundation Systems for Physical AI Deployment USA Today
- Claude Agent for Jira: project-level restriction and per-project GitHub repo mapping?
Claude Agent for Jira: project-level restriction and per-project GitHub repo mapping? Atlassian Community
- 5 Reasons a One-Day Agentforce Operations POC Is Worth Your Time
Most proofs of concept burn months and end in slideware. Agentforce Operations runs the POC in a single day, on us, and ends with two of your real processes already built.
- AI can be an ally in rooting out ransomware threats
AI can be used to prevent cybersecurity threats linked to ransomware, says University of Cincinnati researcher Nelly Elsayed.
Score: 46🌐 MovesJun 26, 2026https://techxplore.com/news/2026-06-ai-ally-rooting-ransomware-threats.html - The AI cold war needs a nonaligned movement
The AI cold war needs a nonaligned movement Nikkei Asia
Score: 46🌐 MovesJun 26, 2026https://asia.nikkei.com/opinion/the-ai-cold-war-needs-a-nonaligned-movement - How AI Can Enhance Content Without Generating It
As much as consumers complain about AI-generated content, advertising experts say AI still has an important place in video creation and production, including for ads. But using AI in content without turning off consumers is a tricky dance. The post How AI Can Enhance Content Without Generating It appeared first on AdExchanger .
Score: 46🌐 MovesJun 26, 2026https://www.adexchanger.com/ctv-roundup/how-ai-can-enhance-content-without-generating-it/ - 3 founders skipped VC funding, used AI to stay lean, and got to $1 million in revenue in year one
3 founders skipped VC funding, used AI to stay lean, and got to $1 million in revenue in year one Business Insider
Score: 45🌐 MovesJun 26, 2026https://www.businessinsider.com/how-outward-intelligence-founders-bootstrapped-seven-figure-revenue-2026-6 - Everyone agrees that you hate AI, but only Mark Cuban sees why Silicon Valley is powerless to fix it
Everyone agrees that you hate AI, but only Mark Cuban sees why Silicon Valley is powerless to fix it Fortune
- AI Made Building Startups Too Easy: Winning Customers Is The Real Challenge
The bottleneck to creating a startup is no longer building the software, but getting people to care.
- [CVPR 2026] An upcycling method for creating true experts within a Mixture-of-Experts structure
[CVPR 2026] An upcycling method for creating true experts within a Mixture-of-Experts structure
- The AI Paradox: How AI is reshaping the commercial office market
Market analysts estimate that this AI-led disruption could eliminate 9-12% of the IT industry revenue in the near term, directly impacting the CRE market
- Alibaba's TRAE Work Adds Design Mode, Closing the Loop From Requirements to Code
Alibaba's TRAE Work platform adds Design mode, enabling full workflow from requirements through design to code generation in a single unified AI development environment.