Tech giants are spending trillions of dollars on AI infrastructure. How the heck will all this be paid for?
The answer depends on whether users, and especially corporate customers, get value from AI tools and services. And, if that happens, will they pay enough to cover this spending spree and leave some left over as profit?
This is the big question that looms over the entire market and economy right now. It’s unnerving to wonder whether there’ll be any real return on investment from this new technology.
I don’t have the answers. However, I recently met an executive who does.
Thomas Bodenski serves as the COO of TS Imagine, which sells software that helps investment firms, banks, and brokers trade, manage portfolios, and monitor risk. He has been in the corporate AI trenches since 2023, using this nascent technology to make TS Imagine more efficient, fast-moving, and ultimately more profitable.
Bodenski’s conclusion so far: The AI spending spree is well worth it, and the ROI he’s seen proves it.
“It’s super beneficial,” he told me in an interview this week. “There’s no way I can go back now. It’s been too successful.”
This is backed up by a recent Wharton study, which found that 74% of enterprises are already generating positive ROI from AI projects.
“This is a pretty big deal, achieving positive ROI is happening faster than I would have expected,” wrote Ethan Mollick, an associate professor of management at Wharton who studies the effects of AI on work, entrepreneurship, and education.
Enter TAIA
TS Imagine’s experience, using data-wrangling AI technology such as Snowflake Intelligence, shows how persistent effort can yield real ROI.
Related stories
Business Insider tells the innovative stories you want to know
Business Insider tells the innovative stories you want to know
Bodenski’s main evidence is TAIA, an AI agent that handles various aspects of customer service and data analysis for TS Imagine.
He shared three specific TAIA examples that he says have eliminated the equivalent of 8.5 full-time employees. TS Imagine employs about 34 data analysts in its data and global client service departments, so that’s significant “person power” gain, Bodenski said.
Customer service
Customer service is already a major area of impact for generative AI in the workplace, and it’s one of the first places where TS Imagine realized benefits.
The firm is constantly bombarded with questions, requests, and rants from demanding, fast-paced clients, including hedge funds, bankers, and traders. It traditionally hired employees to read, classify, and assign these customer-service tickets manually—a stressful task.
Now, TAIA does this automatically, selecting which human professionals are available and best suited to handle each customer request, while adding relevant material to help resolve the issues.
This new AI process takes about one minute per ticket, while the old, manual approach took about 10 minutes. This eliminated the equivalent of three full-time employees, in terms of the time, effort, and other resources required, Bodenski said.
100,000 emails
TS Imagine receives a deluge of emails from numerous vendors who supply and update financial data on a regular basis. It gets roughly 100,000 of these messages every year. Some are incredibly important, while others are irrelevant.
The firm previously tapped financially savvy staff in Hong Kong, London, New York, and Montreal to review every email and either discard it or assign it to the relevant colleagues.
Now, TAIA reads all these emails and automatically classifies and assigns them. TS Imagine gets the same, or better, results from this new AI process at 3% of the cost of the previous manual approach, Bodenski said. Part of those savings comes from the elimination of 2.5 full-time employees, in terms of time and effort, he added.
Corporate actions
TS Imagine helps investors and traders track and understand corporate actions, including stock splits, dividends, and mergers and acquisitions.
The firm often gets updates on corporate actions from exchanges that send the information via PDFs. TS Imagine employees traditionally reviewed this unstructured data and integrated it carefully with customer portfolios.
This stressful and detailed work must get done before trading opens the following day. If an employee gets the math wrong on a stock split, a hedge fund’s investment could swing wildly and inaccurately.
Now, TAIA automatically converts these PDFs into structured data that can be used to update client portfolios and positions more quickly and more accurately, Bodenski said.
“The manual errors have gone,” he added.
This AI automation has saved the equivalent of three full-time employees. TS Imagine still has three other human employees working on this task because it’s so important and complex, although Bodenski is pursuing more AI automation here.
Was it worth all the work?
Through these three specific AI deployments, TS Imagine saved the equivalent of 8.5 full-time employees in terms of effort, time, and other resources required.
Rather than cutting staff, TS Imagine has reassigned employees to more valuable tasks, such as data quality, testing, and client and vendor relationships, Bodenski said.
The firm invested thousands of hours in developing this AI technology. So, was it worth it, in terms of ROI?
“100%,” Bodenski told me. “These large language models are truly revolutionary. They’re not solving things on their own — you have to implement and operationalize the technology properly and make it work for you. But I would not go back. Not possible.”
Sign up for BI’s Tech Memo newsletter here. Reach out to me via email at [email protected].
