First Salesforce, now Workday - tech company after tech company is reducing programmers and increasing sales reps. Both due to AI. Proves what I've been saying all my life: Sales is the smartest, most value added, and least automatable function in a company. No wonder it earns everybody's salary.
Posts by sketharaman
223 publicly visible posts • joined 6 May 2019
Workday erases 8.5% of workforce because of ... AI
DeepSeek stirs intrigue and doubt across the tech world
I registered for DeepSeek 3 days ago. I got a message saying "We're subject to malicious DoS attacks and are going slow on approving new users". I've still not received the email verification code required to login. Then Alibaba announces Qwen that it says outperforms DeepSeek. I'm beginning to wonder if DeepSeek is a serious AI product or a psy op for High Flyer / CCP to short the US markets and rake in a cool half a trillion bucks in one day.
Why does the UK keep getting beaten up by IT suppliers?
Re: "unforeseen technical complexities"
As a supplier for 40 years, RFPs tend to include pretty granular level of details from senior and operating management people. Where things go awry is when expectations fail to materialize during the actual implementation / delivery e.g. CEO wants to change a few things in the new software and, prior to purchase, is confident of getting the operating level people to accept them. But, when the supplier fleshes out the full impact of the change, rubber hits the road. Then there are tons of dependencies with other programs running concurrently elsewhere in the organization. So on and so forth.
Why is Big Tech hellbent on making AI opt-out?
ChatGPT has a Thursday lie down
Capital One two-day outage leaves customers in free-fall
Re: Fail
Reminds of when there was a flood in the West of England and the datacenter of a leading British bank got flooded. When I mentioned that the bank must surely be having sophisticated systems to recover from the flooding, its CIO told me with a deadpan face, "Yeah right, mops and buckets".
Foundation model for tabular data slashes training from hours to seconds
AI frenzy continues as Macquarie commits up to $5B for Applied Digital datacenters
AI hype led to an enterprise datacenter spending binge in 2024 that won't last
Cloud Repatriation Anyone?
Even assuming this analyst is right in its prediction about AI-related infra, not sure if it has considered the effect of Cloud Repatriation in which companies are moving their (AI and non-AI) workloads back to onprem datacenters from public cloud. El Reg had carried an article quoting IDC on this: https://www.theregister.com/2024/10/30/cloud_repatriation_about_specific_workloads/. If this happens at scale, as IDC predicts it would, spends on enterprise datacenters are likely to go up even further rather than down.
Fining Big Tech isn't working. Make them give away illegally trained LLMs as public domain
Are you new here?
Millions of people trained on "In Search of Excellence" and created excellent companies that made humungous profits. How much share of their profits did they give to Tom Peters and Robert H. Waterman?
Billions of people read content on zillions of websites that were accessible publicly and went on to earn diplomas and degrees and get high-paying jobs. How much share of their salaries did they give the publishers of those websites?
Some people argue that it's different in the case of LLMs since they do the slurping and training at unprecedented scale compared to humans but I see no big difference between billions of humans doing something once and one ChatGPT doing that thing billions of times.
So many people and companies have filed so many lawsuits against OpenAI and other GenAI / LLM companies alleging copyright infringement over the past few years. AFAIK not one has received a favorable decision in a court of law. I tend to believe that's because their complaint has no legal merit.
Million GPU clusters, gigawatts of power – the scale of AI defies logic
AWS now renting monster HPE servers, even in clusters of 7,680-vCPUs and 128TB
£1B lawsuit targets Microsoft for allegedly overcharging Windows customers on other clouds
Want to feel old? Excel just entered its 40th year
Inexorable march of progress at SAP threatens to leave users behind
Re: That's enough
Not really. I've been selling IT products and services for four decades. One of the fundamental differences between the two is TELL versus ASK. While a services company can ASK what customers want and scale by simply giving it to them, a product company cannot. To scale in the product business, a product company must have a POV and rally its customers around it, which it does via marketing. Ergo a typical Silicon Valley IT product company spends $2 in Marketing for every $1 in Product / Engineering. I've read so many articles about the disconnect between SAP and its user groups in the last couple of years but I don't recall a single article of that nature in the case of Microsoft, Oracle, Salesforce, and other leading IT product companies during the same period. To me, SAP's inability to rally its customers is Exhibit A of a serious problem in its marketing.
FBI created a cryptocurrency so it could watch it being abused
Workday beats Oracle and Microsoft in UK 'Matrix' ERP deal
Re: "This ERP project is going really well"
Well said! These studies claiming that "early users are struggling to get ROI from AI projects" are daft. I've been selling ERP since 1995. 30 years later, I'll bet that not even 50% of ERP buyers will declare publicly that they get ROI from ERP. As for "ERP project is going well", I'd put that at 25%.
Eric Schmidt: Build more AI datacenters, we aren't going to 'hit climate goals anyway'
Cognizant alleges Infosys swiped its trade secrets
The lawsuit alleges that Infosys built a software by reverse engineering test cases and frontend screens. That's more up the alley of vintage Chinese companies. Last I heard, no Indian IT company is capable of this level of sophistication. Either the lawsuit will be found false and dismissed or it will signify that the Indian IT industry has scaled new heights. Both are good for the Indian IT industry.
India delays planned space station and moon base by five years
Re: comma comment
LOL I thought I had enough local knowledge but I just realized India puts comma after 3 digits up to some numbers and after 2 digits beyond that e.g. 1,000; 10,000; but 1,00,000 and 1,00,00,000. But it shouldn't come as a surprise in a nation full of diversity: Feet for short distance but Meter for intermediate distance and Kilometer for long distance; square feet for apartment size in colloquial context but square meter for the same thing in leave and license agreement; Fahrenheit for body temperature but centigrade for weather.
Under-fire Elon Musk urged to get a grip on X and reality – or resign
Bunch of clowns
X fka Twitter is a private sector company owned by Elon Musk. Any sensible person would realize this and quit Xitter if they don't like it. Only a bunch of commies will expect the owner of a private company to resign. These clowns are the ones who should resign if they're not able to ban Xitter and are still using it instead of moving to another platform.
Patch management still seemingly abysmal because no one wants the job
Re: Let's just ignore the elephant in the room shall we?
Great point! My customer told me that his son's Fortnite on Xbox does 2GB updates every few days. I'm old enough to remember a time when we played computer games off of a 360KB floppy diskette and cannot understand why a video game should occupy 7GB but my customer is a digital native and half my age and even he's speechless why a game update should be 2GB.
Awesome article on a rarely-covered topic. On the back of incidents like CrowdStrike, it's easy to say "all software must have latest updates", "you should not allow vendors to update your software", "you should test all fixes before applying them on production" etc. But this article makes it clear why it's not so easy to follow that advice.
Brace for new complications in big tech takedowns after Supreme Court upended regulatory rules
Chevron Deference
Kudos to SCOTUS for clipping the wings of Sinecure Bureaucrats and Rogue Regulators. While the judiciary also has unelected officials, the big difference between the unelected officials of regulatory agencies and courts is that regulatory agencies can and do poke their nose in business whenever they feel like whereas courts get involved only when there's a dispute.
Payoff from AI projects is 'dismal', biz leaders complain
Re: Actually the most inappropriate applications
Well said. Whether it's telcos or banks or insurers, call centers have been staffed by human CSRs with room temperature IQ for 10+ years. Even seven years ago, I felt that the then dumb chatbots were better than humans for 70% of customer service issues. https://gtm360.com/blog/2017/05/26/can-chatbots-replace-humans/. If anything, that percentage can only go up after ChatGPT / GenAI have entered the scene.
SAP customers may struggle to escape ECC before support shutters if they don't start now
PayPal is planning an ad network built off your purchase history
Re: You just bought a kettle...
Yes, in my company we do digital marketing, and there's overwhelming evidence that targeted advertising has a way higher ROAS compared to traditional "spray and pray" marketing campaigns. If you're seeing ads for kettles after just buying a kettle, the site / app you're on is likely using outdated targeted advertising technology (or switched off targeted ads for you). For 10+ years, leading websites and apps have used cutting-edge targeted advertising technologies to display ads for "related products" i.e. products related to primary product purchased, products purchased by people related to purchaser, etc. To take an example closest to PayPal, many banks in USA have implemented Cardlytics technology to make targeted offers off of credit card purchase history. If your credit card statement for a certain month has many hotel charges, you might receive an ad for "40% discount on your next stay at AirBnB". More at https://gtm360.com/blog/2019/12/13/your-personal-data-is-not-sold-just-used/, https://gtm360.com/blog/2021/12/01/digital-ads-whose-preference-is-it-anyway/, https://gtm360.com/blog/2021/12/15/just-because-you-can-show-digital-ads-doesnt-mean-you-should/.
Google Cloud blunder sinks Australian fund for a week
Financial institutions are well known for being resistant to change and averse to taking on any risk that they can avoid. They have kept their gear onprem for decades. Now, if they're shifting to the cloud without any regulatory mandate and entirely out of free will and volition, it's probably because they're not too convinced about the stability of their sysems when their infra blokes make changes to code in production while under the influence at an Altrincham bar. (Real incident).
If Britain is so bothered by China, why do these .gov.uk sites use Chinese ad brokers?
Kudos!
Great piece! Just when I thought it's fool's errand to expect investigative journalism from mainstream / digital media and have been shilling the Hunterbrook model as future of journalism!!
https://twitter.com/GTM360/status/1775484568428019811
One little-known fact about Real Time Bidding as described in the article: Even the losing RTB bidders get to keep the data received from publisher websites that solicit their bids.
We never agreed to only buy HP ink, say printer owners
Cloud vendor lock-in is shocking, but there's a get out of jail card
Second Sourcing
Great article! TIL "second sourcing". I've been selling IT for nearly 40 years and I'm aware that every other technology introduces some degree of vendor lockin. But I felt cloud raised vendor lockin to the next level since it's the only technology that can literally be turned off by the vendor unilaterally and without any back-and-forth with the customer. I wondered whether customers, including government agencies, who moved to the cloud didn't realize this risk. The first time I read that this is a risk worth taking was in a McKinsey article but it didn't say how it can be mitigated, just that it can. This is the second time I'm reading how the risk can be mitigated. While you yourself admit that the details are sketchy, I do believe that your idea of "second sourcing" sounds promising and worth exploring in futher detail.
UK govt office admits ability to negotiate billions in cloud spending curbed by vendor lock-in
Microsoft unbundling Teams is to appease regulators, not give customers a better deal
Uber Eats to rid itself of pesky human drivers with food delivery by robo Waymo
HPE bakes LLMs into Aruba as AI inches closer to network takeover
AI in Networking
Some form of AI has been around in networking for a long time. In my old company in ca. 1998, we had a software solution that would sniff enterprise networks and use a probabilistic model to detect intrusions and leakages based on anomalies in network behavior. More recently, in ca. 2010, a customer's Network Access Control solution did a similar job in a faster and better (but not cheaper) manner by using an ASIC chip to run the probabilistic model inside of itself instead of relying on the central server's computing power to do so. To be sure, neither of these solutions had machine learning. I'm guessing modern AI will do the same job faster, better and maybe cheaper, and use ML. While that's nothing to be sneezed at, it's unlikely to be transformational for the forseeable future.
How to Netflix Oracle’s blockbuster audit model
Re: Better option
That's always what customers say before they get a feature. Back in the day, many software products had exactly the kind of "technical license limit" that you mention. Customers hated it. They'd complain that, just when they desperately wanted the 31st user to use the software, it was a horrible practice for the software vendor to restrict the use. That's how software vendors progressively withdrew their realtime technical license limits and replaced that with commercial licensing limits, which are enforced only once a year, via audits.
India effectively kills e-wallet used by over 300 million
Not really. PayTM has two operating modes: Wallet and UPI. Only Wallet mode requires topping up and that will not work after whatever date. However, UPI mode uses a bank account as funding source and that bank account can be some other bank than PayTM Payments Bank. Only problem is, PayTM's UPI license is held by PayTM Payments Bank. If that bank is shut down, then PayTM needs to get UPI permit for itself to continue to operate in UPI mode. Not sure whether that will happen or not by the D-Date.
Infosys enjoyed a boom in UK government invoices in 2023
Competition is decreasing in enterprise IT – and you’ll be poorer and dumber for it
Google to bring India’s Unified Payments Interface to the world
US Supreme Court doesn't want to hear Apple, Epic's gripes about in-app purchases
"Apple is allowed to require that in-app purchases must use Apple's own payment method (through which the mega-corp takes a cut of the sales) but it can't stop apps from linking to outside payment systems through which people can also buy stuff (and from which Apple may not be able to take a cut.)".
From the app owner POV:
- First part of sentence means apps MUST use Apple's own payment method.
- Second part of sentence means apps CAN use non-Apple payment method.
Sounds contradictory unless app developers split their price into two parts, route Part A through Apple payment method and Part B through outside payment method.
US tech innovation dreams soured by changed R&D tax laws
I just had a quick glance at 10-Ks of a few American software companies. They show salary costs under either Cost of Revenues or R&D (or both). Most software companies deploy a vast majority of their engineers to develop and maintain their software and a small minority for futuristic R&D work. The new rule only impacts companies that show 100% of salary costs under R&D, which helps companies goose up their Operating Margin (which considers Cost of Revenues but not R&D). What's to stop companies from showing bulk of engineer salary costs under Cost of Revenues (where it rightly belongs in most cases), which can - and is - expensed 100% within the financial year?
How governments become addicted to suppliers like Fujitsu
Uncle Sam will pay for your big ideas to end AI voice-cloning fraud
Artificial intelligence is a liability
Much of customer service is just people blindly following a script anyway.
Well said. Back in 2017, I made the same point in my blog post https://gtm360.com/blog/2017/05/26/can-chatbots-replace-humans/. With ChatGPT / Gen AI taking chatbot capabilities to the next level, AI will likely be better than human CSRs in many more areas. As for displacement of labor, virtually every technology has caused a certain amount of movement of workers from one field to another, I don't think AI will be drastically different. If anything, we might be able to use AI itself to suggest how to redeploy labor displaced by it!