Day two of TechEx North America has been more of a deeper, critical examination of AI in the enterprise, but with a optimistic bent. The AI and Big Data programme opened with reference to what was termed the \u201cAI graveyard\u201d \u2013 that is, AI projects that seem to perform well in pilot, but don\u2019t seem to cut it in the real world. Despite the presence of what might be a negative term, multiple speakers and sessions addressed ways in which the forward-thinking business might not ever have to experience the technological cemetery.
\nThe different show tracks of the second day of this event dived deeper into the pervasive issues that may be affecting AI deployments. Sessions in the Enterprise AI Implementation, ROI and Adoption tracks took stalled pilots as a starting point, and tried to ascertain the reasons behind faltering projects. There was a good deal of sound advice for organisations, with sessions on focusing agentic AI on specific business areas, building agent-ready data foundations (planning for success under the hood), and the realities of token-based AI charging on the business\u2019s finances.
\nAt an infra level, there were deeper discussions too on whether companies should buy or build physical infrastructure for their AI projects, and the best ways to create durable ROI on data and AI projects when all the many effecting factors are given due consideration..
\nIn projects where AI roll-outs get stuck, the core issue could be epitomised by the concept of the \u2018personal copilot\u2019. This works well on a single worker\u2019s desk and for their individual workflows, but doesn\u2019t really scale to a whole department \u2013 never mind a whole business. Many companies report having the budget to start such AI experiments at the level of the single user, and there are usually great results. When said user is a C-suite executive, a personally-achieved efficiency tends to increase the levels of excitement around the company, which has to be considered a positive. But transitioning from this point to meaningful change across the business is where many organisations find their individual struggles and roadblocks. Here was the meat and gravy of day two\u2019s activities on the show floor and the numerous stages at the San Jose McEnery Convention Center.
\nCyber issues
\nDespite the use of terms like \u2018stalled\u2019 and \u2018difficult to scale\u2019, in the Cyber Security and Cloud Expo stage, speakers cited the the speed at which businesses and organisations adopt agentic AI systems as a cause of a \u2018velocity gap\u2019. Where AI deployments are successful, they gain traction fast! But security and governance issues crop up when business units adopt generative AI faster than the security team can govern and ensure the enterprise\u2019s safety.
\nLike the proverbial double-edged sword, AI can be considered as a force that changes and can improve both attack and defence in the cybersecurity space. There are the issues created internally by unbounded agents and large language models, plus the addition to attackers\u2019 arsenals of AI scanning tools that can identify potential exploits.
\nAlso prevalent among the round-table discussions and keynote speeches was the older theme of shadow IT, now presenting in its new guise as shadow AI. If staff place sensitive material into unsanctioned tools for example, or if approved AI systems are poorly bounded and managed, then the attack surface can expand without the cybersecurity team even being aware of it happening. Therefore, data governance and system oversight are becoming more intertwined than before \u2013 this was the message from both cybersecurity strands of the show, and the Cloud and Big Data elements too.
\nFor pure-play cybersecurity functions, zero trust was presented as one answer to the runaway adoption of AI outside the auspices of cybersecurity teams \u2013 the adoption the \u2018denial by default\u2019 position for humans and machines alike. Proof of identity and privilege levels need also to apply to services and agents; that way, automated workflows are subject to the same permission models as every other element in the IT stack.
\nThe second day of TechEx North America was certainly not a rejection of decision-makers\u2019 AI ambitions \u2013 the role of AI and even agents were things of accepted fact among speakers, thought-leaders, and delegates at the event. But there were details and considerations presented by representatives from different industries and business functions, each with positive and insightful things to contribute. Each placed their concerns and their enthusiasms on the table, adding to the discussions around AI implementation in 2026.<\/p>\n
The march of the robots
\nAnd there was a great deal of excitement, still, in many areas of the conference floor. The humanoid robots on show were a source of much enthusiasm (everyone seems to love a lovable android!), but more pragmatically, the new Physical AI track drew some of the show\u2019s biggest audiences. Multiple delegates away from the track cited software coding as the place that has first yielded positive results from the use of large language models in professional settings. And from many places too came the opinion that automated physical systems will be the next industry segment set to benefit from concerted work around new models and their practical harnesses.
\nThe AI models at the heart of next-gen physical AI are unlikely to be LLMs (although these will be useful if the devices are designed to interact with humans), and as such models develop and emerge from their research stages, it\u2019s the TechEx Events series that will be the first to showcase and present these, and how they can work viably in business contexts.
\nNew learning strands to the event
\nThis year\u2019s event saw a welcome injection of pragmatic coding, with hands-on learning sessions that took attendees through spinning up their own AI agentic models, with lessons in how agents can improve themselves, right from interactive Google Colab instances. The TechEx Learning Hub also featured workshops from Nvidia and the ever-popular Google Hackathon, with learners ranging in abilities from those that needed introducing to an IDE through to those that came with software skills already well-tuned. Putting learnings into practice is what this event is all about, whether it\u2019s C-suite decision makers taking on lessons on best strategic practices, or developers turning creative ideas into reality.
\nTechEx takes the cutting edge, and distils it through the business lens; pragmatic yet future facing. Catch the next leg of TechEx in Amsterdam this September \u2013 who knows how far we may have progressed in the space of four short months?
\n(Image source: TechEx Events)<\/p>\n
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and co-located with other leading technology events. Click here for more information.
\nAI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
\nThe post Enterprise AI roadblocks and roadmaps, security and physical AI: Day two at TechEx appeared first on AI News.<\/p>\n","protected":false},"excerpt":{"rendered":"
Day two of TechEx North America has been more of a deeper, critical examination of AI in the enterprise, but with a optimistic bent. The AI and Big Data programme opened with reference to what was termed the \u201cAI graveyard\u201d \u2013 that is, AI projects that seem to perform well in pilot, but don\u2019t seem [\u2026]<\/p>\n
The post Enterprise AI roadblocks and roadmaps, security and physical AI: Day two at TechEx<\/a> appeared first on AI News<\/a>.<\/p>\n<\/div>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"site-container-style":"default","site-container-layout":"default","site-sidebar-layout":"default","disable-article-header":"default","disable-site-header":"default","disable-site-footer":"default","disable-content-area-spacing":"default","footnotes":""},"categories":[1,579,1174,818,1175,499],"tags":[3],"class_list":["post-3216","post","type-post","status-publish","format-standard","hentry","category-ai-and-ml","category-copilot","category-industry-insights","category-physical-ai","category-pilot-programmes","category-techex-events","tag-ai"],"yoast_head":"\n