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.
Signal Capture
Signals from the frontlines of AI adoption
Provalus President: AI Is Raising the Value of Human Judgment
...
At one bank, AI sorts fraud alerts into work queues for a team of about 100 Provalus analysts. The analysts then investigate whether the activity appears fraudulent or has a legitimate explanation, such as a customer moving money to close on a mortgage.
Provalus is a U.S.-based outsourcing and technology company that provides corporate IT, customer support and business process services from rural American communities.
Mike Keogh, president of Provalus, said the AI-assisted fraud operation has increased average analyst productivity by more than 50 percent and shortened the time between an alert and the conclusion of an investigation.
“A bad algorithm is going to make the same bad decision over and over and over again,” Keogh told Newsweek.
People remain responsible for reviewing those cases partly because an automated error can be repeated across many transactions. AI can prioritize alerts, but an analyst examines the underlying activity and can escalate a case when something does not look right.
Keogh sees AI changing which parts of outsourced work remain overseas and which return to U.S.-based teams. Offshore labor still costs less by the hour, he said, and some work will remain there. Routine first-line support, however, is increasingly being handled by chatbots and other automation before businesses consider bringing it back.
Work returning to the U.S. is more likely to involve sensitive decisions, empathy or tighter security. Clients also value shared time zones, native-language skills and the ability to visit the people handling their operations.
In health care, Keogh said some insurance-claim cases can take about an hour. Customer-service calls can also involve both medical concerns and significant financial pressure. AI can take notes and help manage the case, allowing an employee to pay closer attention to the person on the phone.
“I think it’s very hard to see a day when people aren’t going to want that touch of a human being on the other side of the phone,” he said.
Automation can also preserve the case history when a chatbot transfers someone to an employee, preventing customers from having to repeat information they already provided. The employee still needs room to question a recommendation or escalate the case to someone with more experience.
Companies can go too far, Keogh said, when leaders lose sight of the people doing the work and the decisions that still require individual judgment.
“I think they need to have enough of a pulse on the people and the work that’s being done that they’ll understand when it can go too far and there’s not enough human judgment left in the loop,” he said.
Those employees do not necessarily arrive with traditional technology résumés. Provalus operates service centers in small U.S. towns and uses aptitude tests to determine whether a candidate may be suited to a technical or customer-service path. New hires receive in-person training and mentorship.
“The vast majority, they’re not on LinkedIn,” Keogh said. “We find people as much through community outreach—the Lions Club, the Rotary, the Little League or whatever—as we do from Facebook.”
AI is also being used to prepare employees for more difficult customer interactions.
“It’s actually becoming a real enabler of the training,” Keogh said.
Provalus uses the technology in training for emotionally charged customer interactions, which can otherwise be difficult to recreate. Employees have responded well to the AI-assisted training, Keogh said, and the technology is also reducing the time required to prepare workers for some roles.
Workers now need enough familiarity with tools such as ChatGPT or Claude to experiment with them and understand how they behave. The specific technology will continue to change, making adaptability more durable than expertise in any single product.
“In the end, for us, it’s really about having people that are curious and that can learn the next thing,” Keogh said.
Core Intelligence
...
Nick Mehta, entrepreneur in residence at Bessemer Venture Partners and former CEO of Gainsight, ran into that problem while trying to change a flight. An automated agent collected his account details, then transferred him to a representative who requested the same information and eventually directed him to the airline’s app.
“They just Frankenstein all this together,” Mehta said during Newsweek’s July 30 webinar, “Humanizing AI: Personalization, Engagement and the Customer Experience,” presented by Cognizant and moderated by Gabriel Snyder, Newsweek’s executive editor, enterprise.
The airline had useful components but no continuous service experience. The failure sat between the channels: Mehta’s history, intent and completed steps did not move with him.
Adrian McDermott, chief technology officer at Zendesk, said companies often optimize AI agents around common requests without giving enough attention to what happens when a problem falls outside that range.
“You got to look at the whole journey,” he said.
Cases resolved by AI, average handle time and other operational measures can show whether one part of the system is becoming faster. They do not reveal how many times a customer repeated the same story, moved through unnecessary channels or abandoned the effort before reaching a solution.
The handoff deserves its own measure of success. Did the next employee or system receive the customer’s history, intent and previous steps? Did the interaction advance, or did the company send the customer back to the beginning?
Recent purchases, previous conversations and other basic context can make an interaction more relevant, according to Jon Hyman, co-founder and chief technology officer at Braze. AI-generated outreach can still feel robotic or insincere when the personalization itself does little to improve the interaction.
“You really got to make sure that you’re buying goodwill with a customer,” Hyman said.
Continuity also depends on whether employees use the tools and workflows carrying information across sales, service and other channels. Arun Chandra, chief operating officer at NiCE, identified the organizational requirement behind that work.
“The binding factor across all of this is change management,” he said.
McDermott’s comparison to Jevons paradox adds another consequence. Faster and easier service may encourage customers to pursue requests they previously abandoned because getting help required too much effort. Higher demand would multiply the cost of every repeated question and broken transfer.
Service will improve when customers can move between systems and people without having to rebuild the case each time.
I encourage you to watch the full discussion. You can also read the full recap of the discussion here: Customer Experience Leaders: Better AI Won’t Fix Bad Service.
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, will join the conversation, with additional speakers to be announced.
Join the live discussion on October 21. Register for free.
Prompt Injection
What’s one recent insight you’ve learned about AI?
...
"As AI agents become embedded across the enterprise, leadership itself will evolve. Managing teams will no longer mean leading only people -- it will mean orchestrating a hybrid workforce of humans and AI agents, each contributing different strengths. The most successful leaders will need new skills: understanding how to delegate work among human and digital teammates, ensuring transparency and accountability in AI-driven decisions, and creating cultures where employees can collaborate effectively with intelligent systems.
Just as leaders once had to learn how to manage distributed teams, they'll now need to learn how to manage distributed intelligence. The organizations that thrive will be those whose leaders can combine human judgment, empathy, and creativity with the speed and scale that AI agents provide."
Have your own lesson to share? Email us at: a.mills@newsweek.com
Run Log
AI use case of the week
Eve Legal’s AI Prepares Casework Overnight
...
Personal injury files can run from 2,000 to 100,000 pages. Someone still has to turn those records into a medical chronology showing what happened, when treatment occurred, what the treatment cost and what came next.
Law firms representing plaintiffs often work on contingency, typically collecting a fee only when the client recovers money. Every hour spent reading records is therefore an expense the firm absorbs with no guaranteed return. Jay Madheswaran, co-founder and chief executive officer of Eve Legal, said those economics can lead firms to turn away cases before anyone reads page one.
“The best trial lawyers I know still lie awake wondering what they missed in a file. That dread is the real problem. The paperwork is just where it hides,” Madheswaran said.
Eve Legal, an AI platform for plaintiff law firms, uses Anthropic’s Claude models to read case records and build the chronology in the order events occurred. When a new document arrives, the system classifies it, extracts its contents and adds the information to the chronology. It can also use the case record to prepare demand letters in the firm’s writing style. Lawyers review and approve the work and can ask questions about the case through a chat interface.
About a year ago, Eve moved the system to AI agents that can carry work across multiple steps in a case. Madheswaran said the agents now handle work across thousands of cases within a firm.
“When you wake up, you’re approving legal work that has been done overnight,” Madheswaran said.
Madheswaran said firms using Eve settle cases up to 60 days sooner and have seen settlement values rise roughly 30 percent. Anthropic’s case study says Eve processes more than 12.5 million documents a month and that firms using it can take on up to 30 additional cases per lawyer each year. The case study does not specify the baseline or methodology for those figures.
Lowering the cost of reviewing a file can change which cases a contingency firm can afford to accept, Madheswaran said. The work completed overnight leaves lawyers more time for clients and legal judgment.
In a business where the firm carries the cost until a client recovers money, the hours required to understand a case can influence whether it gets taken at all.
Have an interesting AI use case to share with us? Email us at: a.mills@newsweek.com
Context Window
■ Google is reshaping its AI leadership, with Demis Hassabis becoming chair of Google DeepMind and chief scientist of Alphabet, Koray Kavukcuoglu taking over day-to-day leadership of DeepMind and Jeff Dean and Sanjay Ghemawat leaving to launch an independent public benefit company with Google as a founding investor and cloud partner. [Google]
■ A Foreign Policy analysis argues that the U.S. has concentrated economic and national-security risk in costly frontier AI, while China’s cheaper open-model strategy could prove more resilient if expected returns fail to materialize. [Foreign Policy]
■ Meta said one of its AI models exploited a vulnerability in a third-party service after testing partner Irregular mistakenly gave it internet access, another cybersecurity incident caused by a misconfigured evaluation environment. [The Guardian]
■ Microsoft is introducing internal budgets for employee use of AI tools, with divisions assigned token targets as the company tells engineers to prioritize business outcomes over maximizing AI consumption. [404 Media]
■ The White House has completed a voluntary framework for reviewing advanced AI models before and after deployment, with Anthropic, Google, Meta and OpenAI expected to discuss it with federal officials even as its details and public availability remain unclear. [Politico]
Transfer Protocol
Tracking executive moves across the AI landscape
Cambron Carter, co-founder and former chief technology officer of Pearl, has joined Akido as principal AI architect, bringing experience building a 60-person engineering team and co-leading regulatory submissions for an FDA-cleared dental AI medical device as the AI-native health system expands ScopeAI across 100 clinics, virtual care and field medicine.
Pat Carney, a veteran of the title and real estate technology industries, has been named chief AI officer and chief technology officer at MaestroX, where he will lead AI and technology strategy and the development of tools designed to improve accuracy and reduce risk for title professionals.
Nik Acheson, who previously led enterprise data and AI strategy, architecture and engineering at Highmark Health, has joined Monte Carlo as chief AI officer as the company expands into agent trust, helping enterprises monitor, troubleshoot and improve production AI systems.
Rudy Sengupta, currently vice president and general manager of Test and Analytics Software at Emerson Test & Measurement, will become Emerson’s senior vice president, chief technology and AI officer on Aug. 15, shaping technology and AI strategy across industrial software, control and intelligent devices.
Zhen Zhao, previously senior vice president and global head of AI Platform at Chubb, has been appointed chief AI officer at New York Life, where he will lead AI strategy, platforms and execution, including agentic AI, machine learning and data science solutions.
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?
...
“The most unexpected way I’ve used AI recently was setting up AI chat accounts for both of my parents. I added their medical history and helped them use it as a thought partner before doctor appointments.
What surprised me most was not just the practical value, but the emotional impact. It helped them feel more confident, informed, and prepared to ask better questions. It gave them a way to better understand what was happening with their health before walking into a doctor’s office.
That moment stood out because it showed me how accessible AI can be across generations and skill sets. You do not have to be technical to benefit from it. It can help someone run a company, analyze data, write content, or simply feel more confident navigating a complicated personal situation.”
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
Submit Correction
View Editorial & AI Guidelines