By Adam Mills

AI Editor

0

Share

Newsweek is a Trust Project member

See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.

Newsweek

Editor's note: This is AI Impact, Newsweek's weekly newsletter where each week, we will explore how business leaders are unlocking real value through artificial intelligence.

Read More on AI

Tap or click here to get this newsletter delivered to your inbox.

Good morning and thanks for joining me.

“Aspirations are really high. Readiness is low.”

Genpact President and CEO Balkrishan “BK” Kalra used that line near the end of last week’s “AI Impact Forum” webinar, after a conversation about what companies have to fix as they move AI beyond experiments and into everyday operations.

His argument was that the difficult part increasingly sits inside the enterprise. Data may still be fragmented, business processes can vary across teams and markets and employees need enough fluency with the technology to help redesign how work gets done. More capable models do not clear away those problems on their own.

Kalra was especially direct about where companies are now: “Time of proof of concepts is gone. Time of experiments is gone.” He said the next phase is about scale and whether those deployments improve business performance.

I get into that argument in my recap of the webinar (where you can also watch the full discussion), including why Kalra believes data, process and talent debt can constrain what companies are able to do with AI.

This week’s Signal Capture picks up one very specific version of the readiness problem. Companies may already have the customer data they want for an AI initiative and still discover that the permission attached to it does not cover the new use.

Read on for why existing customer permissions can become a problem when AI gives old data a new purpose.

Signal Capture

Signals from the frontlines of AI adoption

Syrenis CEO: Existing Customer Consent May Not Cover New AI Uses

Adam Binks, CEO of Syrenis, which makes enterprise consent and preference management software, has watched companies revisit existing customer permissions as they try to use the same old data in new AI systems.

“I have sat in rooms where a campaign or an AI project stalled for months, not because anyone found a problem, but because nobody could say for certain there was not one,” Binks told Newsweek.

Binks described a retailer that had years of customer contact data and had built an AI-driven outreach model around it. Before launch, the company reviewed permissions collected for earlier campaigns in a handful of markets.

“Nobody had done anything wrong—the permission simply did not stretch that far,” he said.

The retailer went back to customers with a plain-language explanation of the new purpose and relaunched with a smaller audience, which Binks said performed better than the original list would have.

The same mismatch can arise when companies repurpose information for an AI use that falls outside what the customer originally understood and agreed to. The starting point, Binks said, is establishing why the data was collected, what the customer was told at the time and whether a new application is consistent with that permission.

Binks said the difficulty often starts with how enterprise systems were assembled.

“The issue we’re seeing most with larger organizations is that they didn’t design their tech stacks around the customer, but rather designed them around departments, products and individual business problems,” he said.

Marketing, customer service and individual business units may each rely on different platforms and ways of organizing customer data, while acquisitions and third-party providers add more systems that may each appear to hold the definitive record of what the customer chose.

Even the meaning of a preference can vary: Binks said “do not contact” may carry different implications for marketing messages, customer service and regulated communications.

When a customer changes or withdraws permission, the update may have to reach customer relationship management software, marketing platforms, analytics tools, outside processors and other systems using that information.

“AI workflows should check the current consent state before using the data or taking an action, rather than relying on a historic copy of the record,” Binks said.

Another complication is what happens to information that entered a model or training process before a customer changed or withdrew permission. Binks said businesses should be able to trace where the data traveled, which systems used it and whether later customer choices were applied across those systems.

He gave the example of a system predicting that a customer is likely to leave, followed by an AI agent changing an offer, initiating contact, adjusting a service or triggering another process. At that point, Binks said, clear boundaries should define which actions the system can take, what information it can use and when a person must review the decision.

Responsibility for those controls spans functions including privacy, technology and marketing. “What often fails is the space between those functions,” Binks said. “Privacy assumes technology has implemented the rules. Technology assumes marketing understands the permission. Marketing assumes the data it receives is safe to use.”

Binks said leaders often treat consent as something settled when a customer first clicks a button, even though permission is tied to a particular purpose, can change and needs to remain connected to the data as different systems collect, combine and act on it.

People often enter consent discussions expecting to hear what they cannot do with customer data, according to Binks.

“The organizations that get this right end up using more of their data, not less, because they can prove what they are allowed to do and stop arguing about it internally,” he said.

Upcoming Webinars

Overcoming Barriers to AI Transformation in Legacy Industries

Newsweek

AI investment is accelerating, and long-established companies are finding ways to bring the technology into organizations shaped by years of customer relationships, complex operations, critical systems and regulatory responsibilities. The challenge is turning promising pilots into lasting business results without disrupting the strengths that have made those companies successful.

In an upcoming Newsweek AI webinar presented by Cognizant, Gabriel Snyder, Newsweek’s executive editor, enterprise, will moderate a discussion titled “Overcoming Barriers to AI Transformation in Legacy Industries.” The conversation will delve into how established organizations are connecting AI to existing systems and workflows, clarifying ownership and governance and aligning technology investments with people, processes and measurable business outcomes.

Matt Sanchez, chief operating officer at Yahoo, Durga Malladi, executive vice president and general manager of technology planning, edge solutions and data center at Qualcomm and Katy George, corporate vice president of workforce transformation at Microsoft, will join the conversation.

Join the live discussion on October 21. Register for free.

AI Impact Forum Webinar

On November 4, Dr. Ranjit Tinaikar, host of Newsweek’s “AI Impact Forum,” will sit down with Firdaus Bhathena, executive vice president and chief technology and transformation officer at S&P Global, to discuss how large companies are turning advances in AI into business results, and why trusted, differentiated data is becoming increasingly important.

Drawing on Bhathena’s experience at S&P Global, FIS Global and CVS Health, the conversation will examine how organizations make data governed and reusable, redesign workflows around new capabilities and balance speed with trust, security and resilience. They’ll also explore how leaders can keep technology investments focused on customer problems and identify where genuine competitive advantage can emerge.

Join the live discussion Wednesday, November 4, at 11 a.m. ET. Register for free.

Prompt Injection

What’s one recent insight you’ve learned about AI?

...

“Medical devices have always presented a paradox to me. They have been foundational to modern medicine and played a central role in how clinicians diagnose, monitor, and treat disease. And yet, they have not defined the venture category in nearly the same way as biotech or software.

The challenge has not been clinical impact, but how the category fits into the standard venture model. Device companies have historically faced what I think of as an ‘unholy trinity’: long development timelines, regulatory risk, and slow clinical and commercial adoption.

Historically, many medical devices have been discrete, purpose-built clinical instruments built around a specific clinical intervention. Even when the patient impact was enormous, the business opportunity often remained bounded by that initial market.

My ‘aha’ is that AI may be changing the risk-reward equation by changing what a medical device can become. While the hard work of generating evidence, navigating regulation, and earning adoption remains, the best AI-enabled devices can now become clinical intelligence platforms that support earlier detection, novel biomarker discovery, and a more anticipatory model of medicine, meaningfully raising the ceiling for both patient impact and venture-scale outcomes.

It has made me think about devices across three dimensions: capability, setting, and compounding data value.

First, AI can expand a device beyond its initial use case. Once a device is deployed, each additional validated algorithm can broaden both its clinical utility and addressable market across new disease areas

Second, AI can expand where a device can be used. When guidance and interpretation are built into the tool, technologies that once required a specialist can move into primary care and other points of care.

Third, devices can create compounding data value. When used in real-world settings, they can generate datasets that support further validation, the development of new algorithms, and additional clinical use cases.

Optain, an Aegis portco, is one example of this I have seen up close. Its low-cost retinal camera can bring diabetic retinopathy screening into primary care, while additional validated algorithms could eventually help assess broader indicators of vascular, metabolic, and neurological health.

In a resource-constrained healthcare system, the scarce resource is not just capital. It is clinical attention, workflow capacity, and organizational bandwidth for change.

Once a device is integrated into clinical workflow, AI can help that deployment become more valuable over time by supporting additional validated algorithms, broader clinical insights, and greater value for patients and clinicians.”

Have your own lesson to share? Email me at: a.mills@newsweek.com

Run Log

AI use case of the week

MaintainX AI Resurfaces Past Fixes During Equipment Breakdowns

...

Specialized factory equipment can be difficult to repair when documentation is incomplete and the outside teams that installed it have left. Proterra, which makes battery systems for commercial vehicles and equipment, uses MaintainX to manage equipment maintenance at its manufacturing plant in Greer, South Carolina, where it had minimal documentation for some automated machinery.

Nick Haase, co-founder and senior principal at MaintainX, an Autodesk company that makes software for maintenance and operations teams, said Proterra centralized work orders, procedures and screenshots, building its own record of previous repairs and troubleshooting steps.

During a time-sensitive breakdown, technicians can ask MaintainX’s AI to search years of work orders for similar failures, summarize how they were handled and recommend next steps.

“The same model with years of relevant work orders and manufacturing expertise behaves more like an experienced colleague, one who's always available and has seen the failure before,” Haase said.

Haase said the manufacturer reduced downtime by making previous fixes easier to find. The AI can also identify recurring failure patterns that help technicians refine preventive maintenance schedules.

Proterra reported $250,000 in cost savings and more than 192 hours of avoided unplanned downtime from its overall use of MaintainX. Those figures cover the broader maintenance operation, not the AI feature alone.

Over four years, those records grew into a reference library for equipment that arrived with little documentation. And each repair adds another example technicians can retrieve the next time a machine develops a similar problem.

Have an interesting AI use case to share with us? Email me at: a.mills@newsweek.com

Context Window

■ Travel companies are using AI to move trip discovery toward conversational recommendations while also applying the technology to booking, customer service and airport and airline operations. [Newsweek]

■ HeadFirst Global is consolidating 11 businesses under Vertage as it builds a platform designed to help enterprises define work by outcomes and determine the mix of talent and AI agents needed to deliver them. [Vertage]

■ A Cisco and Omdia survey of 1,000 IT and network operations leaders found that 51 percent already run agentic AI that acts in production, while 24 percent are comfortable with AI operating without human oversight. [Cisco]

■ McDonald’s plans to scale its generative AI-enabled ArchIQ ordering system as part of an $8.5 billion, 10-year investment in technology deployment, restaurant modernization and operational efficiency. [CIO Dive]

■ SAP updated its global AI ethics policy to strengthen safeguards, accountability and safety for enterprise AI, with mandatory ethics assessments and human oversight built into its governance process. [SAP]

Transfer Protocol

Tracking executive moves across the AI landscape

Jared Coyle, after serving as chief AI officer for the Americas at SAP, has joined Anaplan as executive vice president, AI, leading its AI Center of Excellence from ideation and roadmap through enterprise delivery and adoption.

Chris Hart, coming from the vice president of data science and AI/ML role at Eli Lilly, has joined ProQR as chief data and AI officer, leading its AI strategy across drug discovery and development for its RNA editing platform.

Florian Quarré, coming from a senior partner and head of digital transformation role at HealthScape Advisors, has joined Premier as chief AI and analytics officer, leading its enterprise AI, data, analytics and technology strategy.

Todd Lohr, KPMG’s national managing principal of clients and markets, will become vice chair and head of Client Technology & Innovation, leading a new group focused on building AI-native businesses and scaling new technology products.

Wei Zheng, who led development of Conductor AI and the company’s broader AI and data strategy as chief product officer, has been promoted to CEO as Conductor makes answer engine optimization the focus of its next chapter.

Know someone on the move in AI? Send job change info to a.mills@newsweek.com

Magic Moment

What’s the most fun or unexpected way you’ve used AI lately?

...

“My CEO stack used to be Claude, ChatGPT, NotebookLM and some Make.com flows. Now it's all custom-built apps in Lovable and Claude Code. ‘Later Gator,’ named after my daughter's favorite book, manages my to-do list and forces a decision on anything untouched after 48 hours. ‘Seldon,’ my virtual second brain, briefs me on my day, drafts my emails, and monitors my most important docs. ‘Murrow’ pulls news from sources I'd never find on my own.

These tools work exactly the way I think. And I built all of them without an engineering background in just a couple of hours. If building your own apps sounds daunting, it isn't. That's actually why we added a Lovable AI lab to MasterClass Executive, our 12-week program built with OpenAI and Chicago Booth. We want every person to leave the program knowing how to build tools like this for themselves.”

Experience some AI magic? Tell us about it at a.mills@newsweek.com

Tap or click here to get this newsletter delivered to your inbox.

View 0 Comments

Request Reprint & Licensing

Submit Correction

View Editorial & AI Guidelines