Cover story – Amid the bursting of the AI bubble, Walmart boasts of its productivity gains through new-generation AI
As the AI bubble debate continues, we see an interesting contrast: companies making the case for AI projects to succeed. According to Stuart’s latest report, Walmart has to be included in the list: A 100x increase in productivity in improving Walmart’s product catalog – the benefits of new generation AI are tangible, says CEO Doug McMillon. Stuart quotes McMillon:
We’re finding concrete ways to use generative AI to improve the experience of customers, members and partners. We’re leveraging data and large language models from others and building our own.
For example, we used generative AI to improve our product catalog. The quality of the data in our catalog impacts nearly everything we do, from helping our customers find and buy what they’re looking for, to how we store inventory across the network, to how we deliver orders. We used multiple Large Language Models to accurately create or improve over 850 million records in a catalog. Without the use of generative AI, this work would have required nearly 100x the current headcount over the same period of time.
But will that be enough to make the online business profitable? Stuart:
Aside from the hype around AI, another technology-related point worth noting is the performance of Walmart’s e-commerce business, which currently has a year-on-year growth rate of 21% globally and 22% in the U.S. That’s far from saying that the online retail arm is making a profit, although global losses are reportedly “continuing to decline.”
And that’s part of the puzzle of generational AI – these aren’t the first productivity gains we’ve seen. But productivity gains and profit/ROI are two different things. As Stuart puts it, if Walmart nails e-commerce, it will likely result in “many questions about whether generative AI is measurable in the end result. But that is for another day.” Indeed – and that’s why there is a difference between the mass market LLM AI investor bubble (which really does exist) and the ROI of AI enterprise projects that, while not yet a clear win, has enough substance to make the AI-for-industry debate complicated/interesting. It’s going to take a while to decipher – and untangle…
Diginomica selection – my top stories on Diginomica this week
Vendor analysis in Diginomica style. Here are my top three picks from our provider coverage:
- The State of Sales – AI is on the way, but Salesforce data says there are limitations – With Dreamforce on board, Stuart analyzes new Salesforce data on the impact of AI on sales teams. Some findings are promising: “83% of sales teams that use AI saw revenue growth in the past year, compared to 66% of teams that don’t. Looking deeper, a third of sales leaders on teams that use AI (33%) said revenue increased significantly in the past 12 months, while half (50%) said it increased somewhat..” But since many non-sales-related activities urgently need to be automated, there is still a lot of work ahead – with the usual obstacles to adoption (expertise, security, data quality, customer (dis)trust).
- Customer-centric innovation – how UiPath wants to shape the future of automation in the public sector – Alyx addresses automation problems in the public sector, with UiPath’s Head of Public Sector (UKI): “(This) also indicates a commitment to ethical customer engagement in the public sector, where the impacts of automation go beyond mere efficiency gains and also impact public trust and social justice.“
- It’s the applications, stupid! C3.ai’s Tom Siebel on the addressable market, deliverable value and AI dogs that don’t hunt – Stuart reports on C3.ai’s progress, with Siebel in an aggressive mood towards the competitors: “There is no AI in this stuff!“C3.ai still has a lot to prove to the market, but amidst the hand-wringing over the AI bubble, it is good to read Siebel’s quote about the impact of more established AI technologies: “Shell has publicly stated that the economic benefit they are getting from C3.ai is over $2 billion a year. Two billion dollars a year, OK? That’s in predictive maintenance. That’s in production optimization in oil wells, process optimization and refining, hydrocarbon loss accounting, and renewable energy integration on a global scale..”
- Celonis and HARMAN are collaborating to generate IP targeting SAP S/4HANA projects – with migration deadlines looming, SAP customers will need all the automation options they can get. Derek reports on a new partnership that aims to facilitate this by sharing co-created IP around “process intelligence” in an SAP environment with generic AI support.
Jon’s lucky bag – Gary documents project results in “Credit card company Yonder uses Zinc to improve its application experience.” Martin goes on a rant in “Monday Morning Moan” about why there are too many VPs for my taste. What’s up with the good old CIO? “Vice president of conference presentations”? Ugh!
Finally, my obsessive research reached a turning point in the question “What does next-generation analyst relations look like – and why should clients care?” I offer you 12 points for your discussion/debate, including:
Vendors boast about “customer success” in their talking points, but they fail to incorporate analyst relationships into that customer success framework. After all, isn’t that how such engagement should be judged? If we’re not making our customers more successful, then what exactly are we doing? No one gives awards for the number of slide decks discussed with analysts in a calendar year.
The best of the enterprise web
My Top Eight
- Social Security Number Hack: National Public Data Confirms Data Theft – There is some disagreement about the extent of this hack, but it is without a doubt a major data theft that involved yet another company you’ve never heard of being entrusted with information that can significantly impact your life – and not protecting that information particularly well. Krebs on Security got to the bottom of the case via NationalPublicData.com. Hack exposes a nation’s data: “Many media outlets falsely reported that the National Public data theft affected 2.9 billion people (this figure actually refers to the number of rows in the leaked records). Troy Hunt of HaveIBeenPwned.com analyzed the leaked data and found it to be a somewhat disparate collection of personal and business data, including the real names, addresses, phone numbers, and Social Security numbers of millions of Americans (both living and deceased), as well as 70 million rows from a database of U.S. criminal records..” A small ray of hope: A significant portion of the disclosed SSNs probably concern deceased (elderly) Americans. Either way – this breach is no joke. Krebs: “Class action lawsuits alone will not lead us to a national debate about what needs to change. Americans currently have very few rights to fight back against the personal and financial surveillance, data collection, and sales that are pervasive in today’s technology-based economy.“
- HOT REVIEW: Beyond the hype: HPE’s Morpheus data acquisition transforms hybrid cloud management – Constellation’s Andy Thurai breaks down an acquisition that went under the radar.
- Changes to controversial California AI safety law fail to satisfy critics – Why didn’t I make this the top story? Because it’s still a moving target, last-minute changes raise questions, and there are smart people on both sides of this debate who know their stuff. Officially, I’m for smart AI regulations – but they need to protect against AI abuse while preserving opportunities for smaller players and open source. Does this law do that? At the time of writing, I’m not 100% sure. Casey Newton at Platformer probably has the most balanced update challenging the law: California’s controversial AI law is close to becoming law, while Gary Marcus makes a passionate case for the law to pass: An open letter to Fei-Fei Li regarding California’s proposed AI regulation SB-1047. We need to follow this one…
- In search of the foolproof AI watermark – On ZDNet, Joe McKendrick covers efforts to combat disinformation and abuse of AI content.
- If only the supply chain were reconfigured – Lora Cecere criticizes sharply: “The evolution of the global supply chain (moving products and materials from one country to another to meet demand) was based on three assumptions: rational government policies, low variability in demand and supply, and availability of materials and transportation. Our technologies and processes have not adapted to the fact that none of these assumptions hold true today..”
touch
It’s hard to top an organization that no one has ever heard of that lost millions of social security numbers to hackers and promises to improve security measures after the fact. Oh, and FlightAware exposed pilot and user information (great job, great job).
Honestly, I don’t have many ideas this week, but I will give out the awards for article title of the week. Tie between “I looked lovingly into the eyes of strangers on ‘Eyechat'” and “Citizens are asked to hand in zombie knives and machetes before the ban.” See you next time…
If you have a #ensw piece that qualifies for hits and misses – for better or for worse – let me know in the comments, Clive (almost always. Most Enterprise Hits and Misses articles are from my curated @jonerpnewsfeed.
