Enterprise hits and misses - AI forces a massive data rethink, Aneel Bhusri returns as Workday CEO, and the AI versus SaaS tension persists

Lead story - AI forces a massive enterprise data rethink - so what's next?

So, Ian unfurled one of the best enterprise AI pieces of the year in Eloi vs Morlocks - does AI prove the spreadsheet rebels were right all along? This is not the type of debate with one right answer, but I'll give it a once-over anyhow. Ian is making a few crucial points: 

• Before companies make a "quality data for AI" push, they might want to rethink the data AI really needs!
• The best AI needs data context. In many cases, that data context is individual and therefore "messy," and outside the scope of transactional systems.
• "Dirty data" is an enterprise reckoning - but you don't need to boil the data ocean to get started.
• The enterprise rebels who rely on their own dirty data shouldn't be punished, but understood (and, maybe, historically appreciated!)

Ian, what say you? 

Most meaningful work in an enterprise rests on information asymmetries and imperfect data. We only know our customers imperfectly. Solving problems means navigating imperfect information. Building new products is basically swimming in imperfect signals across multiple dimensions.

So when we fall for the idea that data has to be perfectly modelled before we can act, we are falling for a fallacy about how our most meaningful decisions are actually made.

I can go with that. However: if most enterprise leaders could snap their fingers today, and have a harmonized, AI-friendly, governed data model across all their applications, they would take it in a second - and worry about the messy data perpetually missing from that model as a separate issue. Obvious gotcha: almost none of them have a comprehensive, real-time AI data stream. Nor do they have the will/desire/budget/skills to do such an overhaul. 

Ian concludes:

The killer ‘data quality’ question for AI, therefore, is not how to build a better enterprise data model, but how to design systems of agency that reflect the implied boundaries and real-time context necessary for enterprise decision-making. Seen this way, the most valuable starting point for enterprise AI data is the messy operational context people have already assembled to make decisions in the real world 

A strong argument, though I'd argue a better place to start is in fact with quality, harmonized data - wherever that might live, along with more conventional AI use cases - and build some momentum (that could include well-labelled unstructured data). But here's where Ian nails it for me: the "messy operational context" is where the real AI action, and biggest gains, will be found. A few more points: 

Ian's argument works best for me when it comes to enterprise decision-making, and/or interacting with an AI "assistant." Yes, you would want to query that assistant on your own personal "dirty" org data, along with more structured enterprise data. However, for agentic/autonomous AI, I would want to see much more reliance on carefully governed data, even in unstructured cases. 

Example: if a service AI agent is responding to a refund request by email, I want that AI agent pulling from the official refund policy, not from an unstructured, "messy" email about returns from a service rep who is might be making an out-of-policy exception for a customer. 

A few more takes/nits:

Ian's point on spreadsheet rebels is fun and vivid, and we all know some. But in many vendor ecosystems - if not most - there are now modern tools that integrate spreadsheets with the financial system of record, often in real-time, so there is an opportunity to balance spreadsheet freedom and organizational discipline (give me a world where neither phrase is a dirty word!). I don't want AI agents working off off rogue spreadsheets. On the other hand, pulling unstructured text documents into decisioning is/should be one of AI's great strengths. 

Not all transactional systems are created equal. Many legacy systems are not even API-friendly, creating more enterprise decision obstacles for AI. Sometimes the best way to advance your operational decision making is to modernize your operational processes and software (yes, that other dirty word, software). 

I believe Ian wants to push organizations to avoid lazy notions of "data quality," and naive pursuits of comprehensive enterprise data models that overlook what we really need for decisions today, tomorrow, this month, this year. 

Maybe talk to someone like Ian before you head down that path... Stay tuned for more diginomica research findings on this topic as we go. For now, check: the diginomica network research - CIOs navigate AI's weight of expectation. 

Diginomica picks - my top stories on diginomica this week

• From chatbot interactions to operational agents - what enterprise deployments reveal about AI readiness today - Alyx did a virtual sit down with Databricks' EMEA CTO, as they hashed out a notable survey on agentic AI: "The practical implication is organizational. AI governance is evolving from a compliance function into an operational capability responsible for observing system behavior, managing risk, and maintaining trust over time. Many enterprises have not yet developed this capability."
• How Colt's CEO Keri Gilder led through a cyber attack while protecting customer networks - Mark Chillingworth shares the story of a trying time for the life of any CEO: "The incident showed the value of constantly looking at the single points of failure in the organization. [Gilder] adds that organizations need to consider how they respond to a single point of failure once an incident happens, highlighting that these incidents are extremely stressful for the staff and that can then lead to burnout and further problems."
• Could medical intelligence re-shape the healthcare industry? AI-for-healthcare is fraught with safety and privacy perils. But is there any industry with more AI upside? George digs in...
• Is agentic commerce enterprise-grade, or not? Yes, says Ayal Karmi - but it needs operational rails - with the emergence of OpenAI and Stripe's ACP protocol, and Google's UCP protocol, agentic commerce has hit a turning point. But what are the pitfalls/opportunities for enterprise brands. Time to revisit my dialogue with nekuda CEO Ayal Karmi. Bonus: I share our agentic commerce autonomy debate as well. 

Vendor analysis, diginomica style. Here's my three top choices from our vendor coverage:

Workday co-founder Aneel Bhusri returns as CEO, ending Carl Eschenbach's 2-year tenure - just another news week in the so-called "SaaSpocalpyse..." Parsing CEO leadership changes almost never boils down to one thing, but Phil has tracked Workday (and SaaS) for a goodly time. He concludes: "Who could possibly have foreseen the extent to which Workday is having to reinvent itself and how it builds and delivers product over the past year or two? The challenge for Bhusri now is to balance that "startup AI founder mindset" with the demands of running and growing an established global corporation." Also see: Phil's fresh Workday use cases, including: Workday's Adam Godson tells us how Paradox re-imagined hiring with AI, slashing time to hire from 21 days to 3 and How AI helps us be more human - Capita's Chief People Officer on its AI journey with Workday. 

Sidenote: this spring, we'll dig into how this will impact Workday's aggressive internal AI development push, with prominent hires from Google, etc. I would bet that Bhusri will double down on this internal cultural/development shift, but we'll get it straight from the source. Also see: Phil's Workday's Adam Godson tells us how Paradox re-imagined hiring with AI, slashing time to hire from 21 days to 3

• HubSpot customer growth soars as multi-hub adoption becomes the norm. But Wall Street's still lashing out with its 'SaaSpocalypse' sulk... HubSpot hit the strong numbers, but as Stuart explains, it was not enough to mollify Wall Street's AI fever dreams. Stuart has the money quote from HubSpot CEO Rangan: "Ownership, accountability and governance, all of these live inside applications, and it's much easier to bring AI into these applications rather than try to extract all of this away as like as if it's just data, it is not."
• Pega targets $2bn revenue milestone as it bets on 'predictable AI' - CEO Alan Trefler slams 'delusional' multi-agent approach - Pega also beat guidance, but as Derek reports, "Shares fell 5% as CEO Alan Trefler doubled down on an AI strategy that prioritizes predictable workflows, calling competitors deploying 'tens of thousands of agents' delusional." I thought Derek did a standout job of contrasting the pros/cons of Pega's agentic approach, noting that there is risk on either the aggressive or conservative side. I'm partial to ambitious AI strategies, but then again, my own research points to the challenges in particular of agent-to-agent communications, where context can get lost. I believe much of it depends on a shared data platform, if so - scaling agents may well be viable. We'll see.  

Enterprise events - use case and analysis - the road shows roll on: 

• Hyperscaler demand for AI infrastructure is off the scale. Good news for Cisco CEO Chuck Robbins - Stuart on Cicso's AI infrastructure re-invention: "I can’t do better than one of my colleagues who summed up Cisco by saying, “You see, there are companies that are making a profit from AI." Also check this use case from Stuart's Cisco Live coverage: Cisco Live Amsterdam - how DHL Express keeps a ThousandEyes focused on its global network to ensure 100% uptime.
• Oracle NetSuite counters the AI investor narrative with its own AI long game - inside NetSuite's AI news with Evan Goldberg - Timed with SuiteWorld New York City, I had a virtual sit down with Evan Goldberg, who made the AI + software case: "Not once in our talk did I hear the usual dreaded "digital worker/teammate" buzzwords. Instead, I heard an emphasis on enhancing the strategic impact of human work, with (human) supervision of AI workflows where needed."

A few more vendor picks, without the quotables:

• Confluent's earnings show solid execution as the company prepares for IBM takeover - Derek
• Bombast or long-term planning? President Harvey Finkelstein makes the pitch that all roads lead to Shopify in the AI shopping era! - Stuart
• Freshworks ends the year in the black as AI adoption takes off and the mid-market prepares to make some fresh choices - Stuart

Jon's grab bag - Dear software leaders, are you having fun yet? Brian would like a word with you: 

I’ve been a broken record at analyst and other vendor events looking for visionary life signs at application software vendors. I must say that I have received considerable amounts of aspirational hyperbole but it actually was more of a short-term wishful thinking regarding AI. It hasn’t been all that original, inspired or worthy of a great stock valuation. [What's a software CEO to do in the current climate? Here’s the urgent 2026 playbook].

Cath kicks off an important new series with What the Google case reveals about the nature and impact of whistleblowing. George examines an intriguing advancement in autonomous vehicles in How generative foundation models are driving autonomous embodied AI. Wayve steers the right route. 

Stuart tees off on the bizarro unsatisfactory moving target of tech regulation in Something for the weekend - quis custodiet ipsos legislatoris? AKA what makes politicians think they’re qualified to regulate tech? Chris unfurled a spicy/visionary three part series on the state of robotics... Start with part one: Robot Futures #1 – why your dirty socks are preventing the future. 

As Chris explains, a huge problem for robots is 3 dimensional training data. What are we doing to solve it? Which approve will get the most traction? Data feedback loops built around simpler real world robotics tasks intrigue me, because you have to start somewhere - but there is plenty more to consider here...

Best of the enterprise web

My top seven

• IBM is tripling the number of Gen Z entry-level jobs after finding the limits of AI adoption - the not-fun entry level job market cannot be reduced to one factor. Yes, AI provokes skills shifts, but as this article notes, it doesn't make younger humans irrelevant:  "'The companies three to five years from now that are going to be the most successful are those companies that doubled down on entry-level hiring in this environment,' Nickle LaMoreaux, IBM’s chief human resources officer, said this week. 'We are tripling our entry-level hiring, and yes, that is for software developers and all these jobs we’re being told AI can do.'"
• AWS and Google Double Down on Cloud and AI: What Enterprise Buyers Need to Know From the Vendors’ Earnings - what should customers expect in their next pricing discussions with the hyperscalers? Brian Undlin from UpperEdge is on the case.
• How to test OpenClaw without giving an autonomous agent shell access to your corporate laptop - hmmm... that sounds like something to avoid, and Louis Columbus explains how.
• Besieged – RedMonk's Stephen O'Grady has one of the highest quality-per-blog-post ratios; this piece on the surging impact of AI on coding is a good example.
• The Algorithmic Bazaar - Thomas Wieberneit has a very well-considered post on the state of agentic commerce: "If your Add to Cart function is tied to a JavaScript button on an HTML page, an AI agent cannot trigger it... You need an API-first architecture where the "channel" is just an implementation detail."
• The Beat Goes On - more tough love for supply chain teams from Lora Cecere.
• Confounders: machine learning’s blindspot - LLMs struggle with causal reasoning, but there are also interesting ways to pursue this. I rarely include vendor material in my picks, but this explanation from Causalens is the most concise I have seen in my causal AI research. And yes, causal graphs can be utilized as part of an LLM's context...
• Bonus video/podcast content: Beating the Odds: AI Leadership, Enterprise Advantage, and Probability Hacking - DisrupTV Ep. 427, Constellation Research. I was on the hot seat for the second half of this show, addressing the paradox of AI leadership. The hosts pushed me to sharpen my take, and that's what live enterprise shows should do...

Whiffs

Microsoft finally takes a breather from their record-breaking whiffs streak... but I have some other dandies: 

No further comment... Moving on: 

Tongue in check of course... 

Now I'm starting to get dizzy... If you find an #ensw piece that qualifies for hits and misses - in a good or bad way - let me know in the comments as Clive (almost) always does. Most Enterprise hits and misses articles are selected from my curated @jonerpnewsfeed.