Utah has authorized a system that can issue initial acne prescriptions without a doctor approving each one in advance. The pilot, developed by New York-based Nolla Health, uses a patient questionnaire and facial scans to select treatments from a short list of physician-approved topical options. The AI cannot invent its own regimens and must hand cases to a licensed physician when it cannot choose confidently. Autonomy is introduced in stages. A physician will approve every prescription for the first 100 patients. For the next phase, covering the first 500 patients, the system issues prescriptions autonomously, and a physician reviews them afterwards each day. Later, doctors will retrospectively review a sample once a week. The service is available through the Nolla Derm app to Utah residents aged 18 and older. Patients complete a 10- to 15-minute intake and facial scan; a treatment plan and prescription can appear within minutes. The pilot starts at $4.99 per month.
A McKinsey Global Institute report projects that artificial intelligence and automation will reshape US jobs by 2035. Demand for about 36 million roles could fall while growth elsewhere creates around 41 million new positions. The core challenge is mobility rather than a shortage of work. In the base case, roughly 11 million workers, about 7 percent of the workforce, or one in 14, may need to leave their current occupations entirely. The range runs from 6 million to 16 million depending on how fast technology is adopted. About 770,000 workers a year may have to switch into completely different occupations, roughly 3.6 times the historical average. Lower-wage workers are 7.6 times more likely than higher-wage workers to face a full career change. Declining demand is concentrated in office support, retail, and transportation, while growth is expected in healthcare, construction, and management.
SAP showcased the Autonomous Enterprise in action at SAP Connect, demonstrating how Joule Work and Joule Assistants generate value across lines of business. CEO Christian Klein said the Autonomous Enterprise has arrived, with Joule and the SAP Autonomous Suite built on the SAP Business AI Platform. Joule Work is beginning customer rollout as an AI interface that lets employees interact with SAP solutions in natural language and complete work across SAP and third-party data, grounded in the SAP Knowledge Graph mapping more than 7 million data fields. Inside SAP, Joule Work is live with 110,000 employees and is associated with about 20% productivity gains in finance, HR, and procurement. Joule Assistants execute operational work within business workflows using established data, authorizations, and audit trails, with customers controlling the level of autonomy. Through the Agent2Agent protocol, Joule can connect with third-party AI and agents.
India’s enterprise AI push is moving beyond experimentation, but the next phase depends less on model availability and more on how organizations manage data, context, security, and costs. That view comes from Elastic’s India leadership as more initiatives shift from pilots toward production-scale deployments. Atul Ahuja, Area Vice President and General Manager for India at Elastic, said several organizations are already tying AI work to specific business outcomes rather than conventional proofs of concept. Enterprises hold data across structured databases, PDFs, scanned documents, video, audio, and legacy records; bringing that information into AI workflows while preserving context remains a major challenge. Elastic argues that search and retrieval across existing sources via connectors and APIs can avoid lengthy consolidation projects. As AI agents proliferate, inference and token costs also rise, making efficient retrieval and context engineering critical to avoid wasteful loops.
ServiceNow has announced AI Workflow Factory and Autonomous Engineer solutions designed to discover where AI can improve work and provide one unified way to build, run, and extend AI workflows across the enterprise. The products aim to address execution gaps caused by fragmented, legacy infrastructure. AI Workflow Factory creates a continuous improvement loop that connects process mining to identify which business processes should change against business KPIs, Autonomous Engineer and Build Agent to help teams build workflow improvements and control quality, and App Engine to run the improved workflows at scale. People set direction and approve outcomes while AI moves more of the work forward, governed by the ServiceNow AI Control Tower. Through Action Fabric, the same governed loop can extend to third-party AI agents and tools. The announcement was made in India at World Forum Mumbai, with partners such as Accenture and Infosys highlighting adoption.
OpenAI will start adding an invisible watermark to text generated by ChatGPT and Codex in the European Union to comply with the EU AI Act. The Act’s transparency rules, which took effect on 2 August, require AI companies to mark AI-generated content so other systems can identify it. The watermark will roll out over the coming weeks to eligible ChatGPT and Codex users on all plans, but only in the EU. Developers using the OpenAI API anywhere in the world can turn it on for select models; it is off by default. OpenAI is not making text watermarking a global default at launch. The method, called textGrain, subtly shapes the model’s word choices to leave a pattern that readers cannot see but a detector can pick up. The signal travels with the text when it is copied. OpenAI said the watermark does not identify the user and did not meaningfully change model performance. Detector access is initially limited to approved researchers and expert organizations.
Tech Mahindra has announced an expansion of its agentic AI readiness through a partnership with Google Cloud focused on Gemini Enterprise. The company is scaling hands-on capabilities across its global workforce to support enterprise demand for production-ready AI. A central element is “Build with Gemini,” a Google Cloud workshop being rolled out to more than 12,500 associates worldwide. Participants learn to create, secure, and govern AI agents using Gemini Enterprise tools, including Agent Runtime, Agent Gateway, and Model Armor, and earn a Google Cloud Skill Badge on completion. Tech Mahindra pairs this trained talent with reusable industry solutions, production deployment experience, and a growing portfolio of agentic AI solutions available through Google Cloud Marketplace. The effort supports its “AI Delivered Right” strategy, which emphasizes responsible, secure, and measurable enterprise outcomes.
A Pittsburgh-based firm, RoboParty, has unveiled RP1 (ROBOTO 01), which it positions as the world’s first high-performance full-stack open-source bipedal humanoid robot. The robot made its global debut at IROS 2026. RP1 is designed for researchers, educators, robotics developers, and embodied AI teams so they can use, modify, train, and extend the system in real development workflows. At the company’s booth, a “Kick Me” demonstration let visitors push or kick the robot; RP1 adjusted its posture and recovered balance under external disturbances. The platform delivers peak joint torque of up to 160 N•m and uses in-house Romomo actuator modules and a real-time motion-control system. Dynamic motion and recovery capabilities are supported by PartyOS, RoboParty’s open R&D foundation that integrates the UFO framework. The project builds on RPO, the company’s earlier fully open-source bipedal humanoid effort.
AI agents that can access files, control apps, and act on behalf of users are prompting operating system makers to rethink permissions. Apple plans additional controls for applications seeking full disk access on macOS, requiring more explicit user action before the permission is granted. The company linked the change to AI agents that can act with less user intervention and to risks that broad access can expose files, mail, messages, and browsing history. Full disk access was designed for cases such as backup software and cannot be obtained silently by an app. Once granted, however, an agent can use the same permission to read information and take further actions. Microsoft is developing agent accounts and agent workspaces on Windows 11 so an agent can run under a separate standard account with limited permissions and a separate environment. It is also working on contained Model Context Protocol servers that run under a separate agent account.
Coforge has introduced an intent engineering framework for enterprises deploying AI agents. The approach adds a layer focused on defining what autonomous systems should optimize for, how performance should be measured, and when they should defer to human oversight. The framework distinguishes three layers of enterprise. AI: prompt engineering, which influences how a model responds; context engineering, which provides information for decisions; and intent engineering, which creates the governance structure for those decisions. Coforge’s method adds objectives, constraints, and measurable outcomes to determine what an agent should pursue. Executive vice president Vic Gupta said most enterprises have invested heavily in teaching AI what to say and what it should know, while the harder challenge is deciding what the agent should optimize for when objectives conflict, risk increases, or exceptions occur.
Utah has authorized a system that can issue initial acne prescriptions without a doctor approving each one in advance. The pilot, developed by New York-based Nolla Health, uses a patient questionnaire and facial scans to select treatments from a short list of physician-approved topical options. The AI cannot invent its own regimens and must hand cases to a licensed physician when it cannot choose confidently. Autonomy is introduced in stages. A physician will approve every prescription for the first 100 patients. For the next phase, covering the first 500 patients, the system issues prescriptions autonomously, and a physician reviews them afterwards each day. Later, doctors will retrospectively review a sample once a week. The service is available through the Nolla Derm app to Utah residents aged 18 and older. Patients complete a 10- to 15-minute intake and facial scan; a treatment plan and prescription can appear within minutes. The pilot starts at $4.99 per month.
A McKinsey Global Institute report projects that artificial intelligence and automation will reshape US jobs by 2035. Demand for about 36 million roles could fall while growth elsewhere creates around 41 million new positions. The core challenge is mobility rather than a shortage of work. In the base case, roughly 11 million workers, about 7 percent of the workforce, or one in 14, may need to leave their current occupations entirely. The range runs from 6 million to 16 million depending on how fast technology is adopted. About 770,000 workers a year may have to switch into completely different occupations, roughly 3.6 times the historical average. Lower-wage workers are 7.6 times more likely than higher-wage workers to face a full career change. Declining demand is concentrated in office support, retail, and transportation, while growth is expected in healthcare, construction, and management.
SAP showcased the Autonomous Enterprise in action at SAP Connect, demonstrating how Joule Work and Joule Assistants generate value across lines of business. CEO Christian Klein said the Autonomous Enterprise has arrived, with Joule and the SAP Autonomous Suite built on the SAP Business AI Platform. Joule Work is beginning customer rollout as an AI interface that lets employees interact with SAP solutions in natural language and complete work across SAP and third-party data, grounded in the SAP Knowledge Graph mapping more than 7 million data fields. Inside SAP, Joule Work is live with 110,000 employees and is associated with about 20% productivity gains in finance, HR, and procurement. Joule Assistants execute operational work within business workflows using established data, authorizations, and audit trails, with customers controlling the level of autonomy. Through the Agent2Agent protocol, Joule can connect with third-party AI and agents.
India’s enterprise AI push is moving beyond experimentation, but the next phase depends less on model availability and more on how organizations manage data, context, security, and costs. That view comes from Elastic’s India leadership as more initiatives shift from pilots toward production-scale deployments. Atul Ahuja, Area Vice President and General Manager for India at Elastic, said several organizations are already tying AI work to specific business outcomes rather than conventional proofs of concept. Enterprises hold data across structured databases, PDFs, scanned documents, video, audio, and legacy records; bringing that information into AI workflows while preserving context remains a major challenge. Elastic argues that search and retrieval across existing sources via connectors and APIs can avoid lengthy consolidation projects. As AI agents proliferate, inference and token costs also rise, making efficient retrieval and context engineering critical to avoid wasteful loops.
ServiceNow has announced AI Workflow Factory and Autonomous Engineer solutions designed to discover where AI can improve work and provide one unified way to build, run, and extend AI workflows across the enterprise. The products aim to address execution gaps caused by fragmented, legacy infrastructure. AI Workflow Factory creates a continuous improvement loop that connects process mining to identify which business processes should change against business KPIs, Autonomous Engineer and Build Agent to help teams build workflow improvements and control quality, and App Engine to run the improved workflows at scale. People set direction and approve outcomes while AI moves more of the work forward, governed by the ServiceNow AI Control Tower. Through Action Fabric, the same governed loop can extend to third-party AI agents and tools. The announcement was made in India at World Forum Mumbai, with partners such as Accenture and Infosys highlighting adoption.
OpenAI will start adding an invisible watermark to text generated by ChatGPT and Codex in the European Union to comply with the EU AI Act. The Act’s transparency rules, which took effect on 2 August, require AI companies to mark AI-generated content so other systems can identify it. The watermark will roll out over the coming weeks to eligible ChatGPT and Codex users on all plans, but only in the EU. Developers using the OpenAI API anywhere in the world can turn it on for select models; it is off by default. OpenAI is not making text watermarking a global default at launch. The method, called textGrain, subtly shapes the model’s word choices to leave a pattern that readers cannot see but a detector can pick up. The signal travels with the text when it is copied. OpenAI said the watermark does not identify the user and did not meaningfully change model performance. Detector access is initially limited to approved researchers and expert organizations.
Tech Mahindra has announced an expansion of its agentic AI readiness through a partnership with Google Cloud focused on Gemini Enterprise. The company is scaling hands-on capabilities across its global workforce to support enterprise demand for production-ready AI. A central element is “Build with Gemini,” a Google Cloud workshop being rolled out to more than 12,500 associates worldwide. Participants learn to create, secure, and govern AI agents using Gemini Enterprise tools, including Agent Runtime, Agent Gateway, and Model Armor, and earn a Google Cloud Skill Badge on completion. Tech Mahindra pairs this trained talent with reusable industry solutions, production deployment experience, and a growing portfolio of agentic AI solutions available through Google Cloud Marketplace. The effort supports its “AI Delivered Right” strategy, which emphasizes responsible, secure, and measurable enterprise outcomes.
A Pittsburgh-based firm, RoboParty, has unveiled RP1 (ROBOTO 01), which it positions as the world’s first high-performance full-stack open-source bipedal humanoid robot. The robot made its global debut at IROS 2026. RP1 is designed for researchers, educators, robotics developers, and embodied AI teams so they can use, modify, train, and extend the system in real development workflows. At the company’s booth, a “Kick Me” demonstration let visitors push or kick the robot; RP1 adjusted its posture and recovered balance under external disturbances. The platform delivers peak joint torque of up to 160 N•m and uses in-house Romomo actuator modules and a real-time motion-control system. Dynamic motion and recovery capabilities are supported by PartyOS, RoboParty’s open R&D foundation that integrates the UFO framework. The project builds on RPO, the company’s earlier fully open-source bipedal humanoid effort.
AI agents that can access files, control apps, and act on behalf of users are prompting operating system makers to rethink permissions. Apple plans additional controls for applications seeking full disk access on macOS, requiring more explicit user action before the permission is granted. The company linked the change to AI agents that can act with less user intervention and to risks that broad access can expose files, mail, messages, and browsing history. Full disk access was designed for cases such as backup software and cannot be obtained silently by an app. Once granted, however, an agent can use the same permission to read information and take further actions. Microsoft is developing agent accounts and agent workspaces on Windows 11 so an agent can run under a separate standard account with limited permissions and a separate environment. It is also working on contained Model Context Protocol servers that run under a separate agent account.
Coforge has introduced an intent engineering framework for enterprises deploying AI agents. The approach adds a layer focused on defining what autonomous systems should optimize for, how performance should be measured, and when they should defer to human oversight. The framework distinguishes three layers of enterprise. AI: prompt engineering, which influences how a model responds; context engineering, which provides information for decisions; and intent engineering, which creates the governance structure for those decisions. Coforge’s method adds objectives, constraints, and measurable outcomes to determine what an agent should pursue. Executive vice president Vic Gupta said most enterprises have invested heavily in teaching AI what to say and what it should know, while the harder challenge is deciding what the agent should optimize for when objectives conflict, risk increases, or exceptions occur.