Enterprise hits and misses - AI agents get a deterministic reality check, and the SaaSpocalypse hype gets new critics

Lead story - Is the SaaSpocalypse a real thing? No, says... Geoffrey Moore? 

We've published a slew of explanations on why agentic AI isn't going to eat SaaS. But I didn't expect one of the best to come from Geoffrey "Crossing the Chasm" Moore. But check Stuart's Can agentic AI 'cross the chasm' without falling in? Tech thought leader Geoffrey Moore assesses progress in straddling the divide. Stuart's virtual sit down with Moore produced this money quote: 

SaaS contains almost a half a century of business acumen, and I'm sorry, but you're not going to just displace a half a century of experience, you're just not. Every company in the world runs on systems of record and has overlaid systems of engagement, so nobody's going to rip them out...If you believe that the end product is a deterministic action, the universe’s  deterministic actions live in the systems of record and systems of engagement. They don't live in AI. OpenAI has no library of determinist actions. SAP does, Oracle does, Hubspot does, Salesforce does, ServiceNow does.

Yes, that's the crux of it, the deterministic essence. With smaller models and superior context, you can improve (probabilistic) LLM accuracy. But a library of deterministic workflows? Not so fast - Moore is right. 

Beyond workflows, there is the separate matter of decision intelligence using LLM agents/assistants. Yes, you could make those decisions on top of a harmonized data layer across a range of SaaS applications, abstracting away from the SaaS vendor in question. It will be interesting to see who wins that business. Software vendors will argue for embedded intelligence and rich semantics, e.g. SAP's partnership with Databricks - making the case that this is superior to agentic best-of-breed API data calls. We'll find out.

Stuart hits home again via SaaSpocalypse No! Stop with the rumors of software's death and listen to Box CEO Aaron Levie. He quotes Levie: 

I'm sympathetic to the idea that you're going to see more competition in software, and I'm sympathetic that agents will do some capabilities that we've currently had in software. Those two things have to happen. They have to be inevitable.

The part that I'm not sympathetic to is actually the two parts are not synthetic. One is that we're just going to code our own CRM system or ERP system. I largely don't think that's going to happen. We all collectively have a fixed amount of resources, and we can deliver more things that [our customers] want to buy and better sort of support their needs. Coding an ERP system is not on that list of things our customers cares about.

After one-too-many LinkedIn skirmishes, Brian Sommer lured me into a blown gasket on this same topic, which we released as a podcast: ERP vendors versus AI agent disruptions - an informal discussion gets heated.

• Despite some of the over-the-top AI happy talk and investor infatuations, market pressure on incumbent software vendors is a good thing, for vendors and customers.
• Not all Saas (or ERP) is created equal. Modern solutions that flexibly adapt business models to the pace of today's economy are pretty rare; SaaS software with innovative pricing models are rare indeed. (See more on this in the best-of-web section).
• ERP and CRM as categories are getting long in the tooth; the need for deterministic workflows doesn't mean software categories themselves will live on forever. End-to-end data/process platforms are more in tune with where we are headed.
• Tacking best-of-breed agents on whatever software you can find and declaring that 'software is dead' is a recipe for governance and orchestration roadblocks.

As Stuart concludes from his talk with Moore: 

I felt coming into 2026 that this would be a year of not so much correction as adjustment of expectations around AI and agentic AI in particular and that this was going to take some modified messaging on the part of vendors to set user expectations accordingly without at the same time putting them off of the tech’s potential.

A slew of keynotes are being planned right now; we'll soon find out where that messaging lands. 

Investors balk at AI infrastructure spend - Wall Street loves AI (well, mostly, except for that pesky ROI thing), but doesn't like SaaS that actually works... Investing in AI infrastructure? Stuart tries to find the market wisdom amidst the fickle: 

• Amazon raises AI-driven CapEx spend by 50% this year; Wall Street stays calm - NOT!!! - Stuart: "Totting up some of the big tech spending plans to date, if we combine the estimated outlay of Amazon, Microsoft, Google and Meta combined that comes to a collective  sum of more than $630 billion this year. Will that be enough to convince the ‘show us the AI money’ red braces brigade that this is the direction of travel and to get their heads around the idea? No, of course it won’t as the sliding Amazon stock price showed." Stuart's vigil didn't end there:
• Wall Street has another funny turn over Alphabet's plans to ramp up spending to further boost cloud successes. Who is going to teach the red braces brigade about the math of investing in the future? 

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

• Exclusive - new Qualtrics CEO Jason Maynard on why ‘experience context’ will make or break agentic AI - Derek has tracked Qualtrics closely for years.... So what does he make of his first sit down with the new CEO - and will "systems of decisions" and "experience agents" stick?
• Atlassian hits $1 billion quarterly cloud revenue - and here's why it thinks AI creates more work, not less - Alyx on Atlassian's earnings milestone, and field lessons on AI coding and collaboration.
• Research - the barrier to AI has shifted from 'should we?' to 'why isn't it working?' - Derek parses fresh data from Celonis: "It's encouraging, despite mixed successes, that we are moving beyond ‘AI is the solution to everything’ towards ‘let’s get the fundamentals right.'"
• Determining an agentic future - why Salesforce is evolving its AI thinking beyond the limitations of the LLM - Stuart on Salesforce's LLM maturity lessons: "As an AI provider you need to provide guarantees that these can be met. Salesforce is approaching this by blending deterministic and non-deterministic tech."

Event wraps come in - with more events on deck. Before long, the event calendar will get wacky... for now, cold fronts, storms nobody asked for, and time for event wraps + standout use cases: 

• Dynatrace Perform 2026 - how two organizations transformed business observability into strategic advantage - Alyx
• Acumatica Summit 2025 - SMB customers air out their views on AI, automation, and data quality - Jon

A few more vendor picks, without the quotables:

• Snowflake Build 2026 - the race to prove AI investments aren't wasted money - Mark Samuels
• Freshworks is capturing the mid-market - but can it evolve from workflow automation to AI governance? - Derek
• Bringing a B2C mindset to a B2B challenge - Piyush Sagar Mishra, Twilio Head of Marketing Innovation and AI, explains how - Barb
• How Vodafone is taking the lead for hyperscale IoT. (Spoiler - simplicity is the key, says the CEO... - George

Jon's grab bag -  Cath hits on key issues that matter all year round in Black History Month - how to survive and thrive as a Black woman in tech. Madeline has one of my fave tech stories of the year in Tech for good - the death of animal testing? Tech advances might render it obsolete... Phil shares the view from a new workplace startup that is, appropriately enough, founded by former engineers on Meta's underrated (and unceremoniously kiboshed) Workplace product: Former Workplace from Facebook engineers launch AI start-up Slashwork to banish teamwork slop. 

Katy nails down a crucial Ai productivity insight in Standing at the crossroads - PwC on the difficulties of getting productivity gains from Copilot. Quoting PwC's Stephanie Mosticchio: 

Would I use an AI agent to email execs at Microsoft? Do I want it to send this autonomously? No, I do not. For productivity gains you need to look at AI agents in supply chain apps or predictive maintenance apps.

Bingo indeed. I haven't heard rave reviews of Copilot, but we can expect these kinds of productivity apps to - at best - scratch the surface, compared to the kinds of meaty use cases Mosticchio cited. 

Stuart spices up a stellar week with Haves and have nots, Europe's big problems, and why Palantir is too busy in the US to help much - CEO Alex Karp gets bombastic with the tough love. Once more from Stuart: NewsCorp isn't messing around with content theft: News Corp's ultimatum to AI firms - fund fecundity and get sued or get with the program and do a content deal:

I like Thomson’s turn of phrase when he warns that anyone who doesn’t get the company’s message around AI and take it on board will be demonstrating a policy decision “to fund fecundity will mean that AI stands for Artificial Intransigence.'

Yep, that's a blammo, and, a wake-up call for Big AI execs, at least those who haven't made content deals with News Corp.

Best of the enterprise web

My top six

• SaaS death knell storyline illogical, but margin compression is here - Constellation's Larry Dignan issued a must-read piece of his own on the SaaS versus agentic AI face-off. One crucial point: SaaS has been under margin pressure, and rightly so. On my DisruptTV appearance on the paradox of AI leadership, I argued that AI has an uncanny way of disrupting stale business models - models that should have self-disrupted years ago (e.g. research, consulting). Though SaaS is not as disrupted as research and consulting (yet), the same applies...
• MCP shipped without authentication. Clawdbot shows why that's a problem - Louis Columbus on the flaws of rolling out MCP out-of-the-box: "When Knostic scanned the internet, they found 1,862 MCP servers exposed with no authentication. They tested 119. Every server responded without requiring credentials."
• US companies accused of ‘AI washing’ in citing artificial intelligence for job losses - Reasons for layoffs are more complex than facile tech efficiency claims. My current read on this: most of the jobs supposedly "lost" from AI came either prematurely, or due to related AI expenses taking cash oxygen, e.g. data center buildouts (alongside the other economic factors cited in this piece). However, I do believe some teams have been able to slow hiring due to so-called improved cognitive load due to AI-adapted workflows. It is most helpful to analyze each company's real gains and losses here, but we can only do that with more transparent discussions on labor versus automation.
• The Compute Factor: Sam Altman Might Be Thinking 10GW Priced at $100 Billion Might Not Be the Best Choice - an interesting analysis of the NVIDIA and OpenAI PR skirmishes - a topic with real world market implications.
• AI Slopageddon and the OSS Maintainers – a well-thought piece on the impact of AI code slop on open source by RedMonk's Kate Holterhoff.
• Failure: a Building Block for Success – Lora Cecere has some more tough love for today's supply chain leaders. 

Whiffs

Things that make you go...

Not everything online is FOMO: 

Microsoft, meanwhile, is on an epic Hits/Misses hot streak: 

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.