This bank is using AI versions of its analysts to meet clients' demand for videos
Bank of America uses " Banker Assist" to aggregate information to offer insights unique to each client, while Goldman Sachs has a " GS AI Assistant" that functions as an in-house ChatGPT for staff.
Swiss bank UBS has gone further, using AI to generate avatars of analysts that explain their research to clients, and it's planning to do this more.
The Swiss bank started using AI avatars of some analysts in January. About 36 UBS analysts, or 5% of its total, have volunteered to take part. They cover sectors including technology, consumer goods, and energy.
UBS's use of AI avatars was first reported by The Financial Times.
Using OpenAI and Synthesia tools, a script is generated in a matter of seconds that is then edited by staff.
Scott Solomon, head of global research technology at UBS, told Business Insider that his team started creating videos of analysts a decade ago, but capacity restrictions meant they were capped at about 1,000 annually.
Analysts were writing an average of two notes a week but would only go to the video studio once a quarter, he said.
The new tools are "enabling somebody to use a capability in video that they weren't really able to use before," he said.
It also gives clients another way to digest information and meet their rising demand for video, Solomon said.
He compared an avatar to other parts of an analyst's toolkit. "When an analyst joins UBS, we give them Excel, we give them our authoring platform, we give them a CRM [customer relationship management] tool so they can talk to clients. I want them to have an avatar," he said.
Solomon said the next step would be integrating the technology so that a video can eventually be created when an analyst publishes a note — without the need for editing.
He said he hoped this would become possible by the end of the year.
Even if the process was fully automated, UBS said analysts will still assess a video based on their notes before it is sent to clients.
Solomon said that ideally, the avatars would eventually become part of the onboarding process, so that whenever a note is published, there's a video too.
The next step would be integrating this capability directly into the authoring platform.
"We have the script generator, we have the ability to send the script to generate the avatar, and then we obviously have the ability to deliver the avatar to clients," Solomon said.
"We want to string all that together so that as they're writing the note, they can get the video with it as well. Our goal is absolutely not to do 50,000 videos a year, but clearly there's an opportunity to do more videos than we are today."
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
an hour ago
- Yahoo
Inside the Secret Meeting Where Mathematicians Struggled to Outsmart AI
On a weekend in mid-May, a clandestine mathematical conclave convened. Thirty of the world's most renowned mathematicians traveled to Berkeley, Calif., with some coming from as far away as the U.K. The group's members faced off in a showdown with a 'reasoning' chatbot that was tasked with solving problems they had devised to test its mathematical mettle. After throwing professor-level questions at the bot for two days, the researchers were stunned to discover it was capable of answering some of the world's hardest solvable problems. 'I have colleagues who literally said these models are approaching mathematical genius,' says Ken Ono, a mathematician at the University of Virginia and a leader and judge at the meeting. The chatbot in question is powered by o4-mini, a so-called reasoning large language model (LLM). It was trained by OpenAI to be capable of making highly intricate deductions. Google's equivalent, Gemini 2.5 Flash, has similar abilities. Like the LLMs that powered earlier versions of ChatGPT, o4-mini learns to predict the next word in a sequence. Compared with those earlier LLMs, however, o4-mini and its equivalents are lighter-weight, more nimble models that train on specialized datasets with stronger reinforcement from humans. The approach leads to a chatbot capable of diving much deeper into complex problems in math than traditional LLMs. To track the progress of o4-mini, OpenAI previously tasked Epoch AI, a nonprofit that benchmarks LLMs, to come up with 300 math questions whose solutions had not yet been published. Even traditional LLMs can correctly answer many complicated math questions. Yet when Epoch AI asked several such models these questions, which were dissimilar to those they had been trained on, the most successful were able to solve less than 2 percent, showing these LLMs lacked the ability to reason. But o4-mini would prove to be very different. [Sign up for Today in Science, a free daily newsletter] Epoch AI hired Elliot Glazer, who had recently finished his math Ph.D., to join the new collaboration for the benchmark, dubbed FrontierMath, in September 2024. The project collected novel questions over varying tiers of difficulty, with the first three tiers covering undergraduate-, graduate- and research-level challenges. By February 2025, Glazer found that o4-mini could solve around 20 percent of the questions. He then moved on to a fourth tier: 100 questions that would be challenging even for an academic mathematician. Only a small group of people in the world would be capable of developing such questions, let alone answering them. The mathematicians who participated had to sign a nondisclosure agreement requiring them to communicate solely via the messaging app Signal. Other forms of contact, such as traditional e-mail, could potentially be scanned by an LLM and inadvertently train it, thereby contaminating the dataset. The group made slow, steady progress in finding questions. But Glazer wanted to speed things up, so Epoch AI hosted the in-person meeting on Saturday, May 17, and Sunday, May 18. There, the participants would finalize the final batch of challenge questions. Ono split the 30 attendees into groups of six. For two days, the academics competed against themselves to devise problems that they could solve but would trip up the AI reasoning bot. Each problem the o4-mini couldn't solve would garner the mathematician who came up with it a $7,500 reward. By the end of that Saturday night, Ono was frustrated with the bot, whose unexpected mathematical prowess was foiling the group's progress. 'I came up with a problem which experts in my field would recognize as an open question in number theory—a good Ph.D.-level problem,' he says. He asked o4-mini to solve the question. Over the next 10 minutes, Ono watched in stunned silence as the bot unfurled a solution in real time, showing its reasoning process along the way. The bot spent the first two minutes finding and mastering the related literature in the field. Then it wrote on the screen that it wanted to try solving a simpler 'toy' version of the question first in order to learn. A few minutes later, it wrote that it was finally prepared to solve the more difficult problem. Five minutes after that, o4-mini presented a correct but sassy solution. 'It was starting to get really cheeky,' says Ono, who is also a freelance mathematical consultant for Epoch AI. 'And at the end, it says, 'No citation necessary because the mystery number was computed by me!'' Defeated, Ono jumped onto Signal early that Sunday morning and alerted the rest of the participants. 'I was not prepared to be contending with an LLM like this,' he says, 'I've never seen that kind of reasoning before in models. That's what a scientist does. That's frightening.' Although the group did eventually succeed in finding 10 questions that stymied the bot, the researchers were astonished by how far AI had progressed in the span of one year. Ono likened it to working with a 'strong collaborator.' Yang Hui He, a mathematician at the London Institute for Mathematical Sciences and an early pioneer of using AI in math, says, 'This is what a very, very good graduate student would be doing—in fact, more.' The bot was also much faster than a professional mathematician, taking mere minutes to do what it would take such a human expert weeks or months to complete. While sparring with o4-mini was thrilling, its progress was also alarming. Ono and He express concern that the o4-mini's results might be trusted too much. 'There's proof by induction, proof by contradiction, and then proof by intimidation,' He says. 'If you say something with enough authority, people just get scared. I think o4-mini has mastered proof by intimidation; it says everything with so much confidence.' By the end of the meeting, the group started to consider what the future might look like for mathematicians. Discussions turned to the inevitable 'tier five'—questions that even the best mathematicians couldn't solve. If AI reaches that level, the role of mathematicians would undergo a sharp change. For instance, mathematicians may shift to simply posing questions and interacting with reasoning-bots to help them discover new mathematical truths, much the same as a professor does with graduate students. As such, Ono predicts that nurturing creativity in higher education will be a key in keeping mathematics going for future generations. 'I've been telling my colleagues that it's a grave mistake to say that generalized artificial intelligence will never come, [that] it's just a computer,' Ono says. 'I don't want to add to the hysteria, but in many ways these large language models are already outperforming most of our best graduate students in the world.'
Business Insider
2 hours ago
- Business Insider
REV Group price target raised to $34 from $30 at Goldman Sachs
Goldman Sachs analyst Jerry Revich raised the firm's price target on REV Group (REVG) to $34 from $30 but keeps a Sell rating on the shares. The company's Q2 results revealed higher Fire business throughput and cost control amid strong production performance in Specialty Vehicles, the analyst tells investors in a research note. The firm adds however that REV's Recreation segment EBITDA was only in-line, with lower margins driven by lower units. Confident Investing Starts Here:
Yahoo
3 hours ago
- Yahoo
After its data was wiped, KiranaPro's co-founder cannot rule out an external hack
Indian grocery delivery startup KiranaPro's recent data loss story has more holes than Swiss cheese, as the startup remains unclear whether the incident was an internal breach or an external hack. Last week, the Bengaluru-based startup discovered that it could not access its back-end servers and that all its data, including its app code, had been deleted from GitHub. The startup on Friday blamed a former employee for the breach. However, in an interview, KiranaPro co-founder and CEO Deepak Ravindran conceded that the company had not deactivated the employee's account after they departed the company and cannot rule out the possibility of subsequent malicious misuse of their account. "If we go deeper, we have to do a real forensic investigation. We are going to talk [about] this with our board, the investors, and we are going to get a formal opinion on that also with our legal advisers," Ravindran told TechCrunch. Earlier on Friday, Ravindran claimed in a post on X that the incident that affected its data was an internal breach. "After careful investigation, we conclude that this was not a hack. No external party penetrated our ordering or payment systems, exploited vulnerabilities, or bypassed security protocols," he wrote. The co-founder also explicitly shared a screenshot of a LinkedIn profile of one of KiranaPro's former employees on X on Thursday, alleging that they had deleted the startup's code. (TechCrunch is not sharing the post's link, as the startup has yet to offer concrete proof supporting its position.) "[T]his was an internal data breach. Specifically, it was the result of actions taken by a trusted internal employee who had legitimate access to our systems," the co-founder wrote in his post on Friday. "This individual intentionally deleted critical server logs while they were being tested and/or edited, an action that goes directly against our policies, our principles, and the trust we place in our team." When TechCrunch asked if KiranaPro could rule out whether any third party had maliciously gained access to the former employee's account, Ravindran could not. "We have to do a complete forensic check on the company. We have to do the entire IP scan. We have to look at where the tracks happened. We have to check the computers, MacBooks, and whatever is used. Everything has to be done. Then we have to spend money … so, that's why we decided not to," he told TechCrunch. Then what was the basis of Ravindran's allegation? It was a GitHub response, a copy of which he shared with TechCrunch. The response included a username, which Ravindran said was associated with the former employee. "All we have is the emails that we got from GitHub, stating that [the former employee's username] as an individual is the one who deleted the account. We haven't done the investigation further," Ravindran told TechCrunch. Launched in late 2024, KiranaPro operates as a buyer app on the Indian government's Open Network for Digital Commerce. The startup allows more than 55,000 customers in 50 cities to purchase groceries from their local shops and nearby supermarkets using its voice-based interface. The company also supports local language inputs, including English, Hindi, Malayalam, and Tamil. Ravindran stated that they decided to call out the former employee based on the company's "belief system," as they claim the former employee deleted the data after their sudden termination. However, the startup said it is not aware if there were enough protections on the former employee's devices, such as multi-factor authentication, to restrict malicious third-party access, like malware. The company confirmed it did not remove the employee's access to its data and GitHub account following his departure. "Employee offboarding was not being handled properly because there was no full-time HR," KiranaPro's chief technology officer, Saurav Kumar, confirmed to TechCrunch. Alongside its code saved in GitHub, KiranaPro also lost access to its Amazon Web Services (AWS) account, which included its customer data and their transaction details. Ravindran told TechCrunch that the GitHub data was restored after getting its backup from one of their employees. The startup also regained access to its AWS account along with its customer data. Both the co-founder and CTO said the AWS account was protected by multi-factor authentication, but neither could say how the account was accessed, as nobody else had physical access to Ravindran's phone, which generates the multi-factor code. Nonetheless, Ravindran claimed that the customer data stored in the AWS cloud remained intact and was not accessed by any third parties, nor was it downloaded by the former employee in question. "Because if that is the case, I will get its notification on email or anything [sic]," he said. That said, Ravindran stated that the startup has enough evidence to file a formal complaint with the police, but said that its investigation is ongoing. The startup has also not fully paid its current employees, the company's co-founder confirmed, soon after the company raised a seed round of ₹100 million Indian rupees (about $1.2 million), which Ravindran said has yet to be fully wired. The startup counts Blume Ventures, Unpopular Ventures, and Turbostart among its institutional venture backers, as well as Olympic medalist PV Sindhu and Boston Consulting Group managing director Vikas Taneja among its angel investors. It has 15 employees located in Bengaluru and Kerala. Sign in to access your portfolio
