I attached below one of the first stories of the "rotation" turning back from the WSJ today.

It is quite amusing watching the writers of the narratives seizing on the smallest of facts and developments, and then spinning them into some big story about how the world will evolve.

For example, yesterday there was a story about how small businesses are using agentic AI firms to vibe-code their own CRM systems to overcome limitations with the Saleforce product. Yes, SMB owners actually getting AI-native vibe coders to help them spin up new systems from scratch, when there is a host of SaaS CRM offerings already available in the market. So, on a day when SaaS generally was rebounding, $CRM was down around 1%, and today it was up 4.3% as it cannot resist the rest of the SaaS "swing back to software" ... or ,... perhaps it is just another dead cat bounce? (cue technical analysts)

My current thinking is that it really does look like some of the extreme P/Es of the high growth SaaS firms have been reset, and I think this is structural and probably justified if you look back at P/Es over the last 3-5 years.

So where to from here? As @Solvetheriddle says, for long term investors what matters is "it all depends on the trajectory of LT earnings." AI and its rollout is going to generate "waves" for the forseeable future, and over time we'll see which firms will stay afloat, and perhaps even surf the waves to race ahead.

I'm not one to "call the bottom", but I think a year or two down the track we'll look back at today and see that it was a great time to pick the winners and load up on them.

Remember, in the article below, the timeframe is 4 days. 4 days! I'm more interested in looking at trends over 4 years.

Hi @mikebrisy

Agree with your how ridiculous some of the narrative-spinning that goes on in the press these days - spin a story out of whatever micro development has popped up in the moment to suit the story-du-jour.

Also very much agree with you regarding the potentially structural re-set in SaaS PEs. Over the last several years the market rewarded them with ever-more lofty PEs, likely a combination of re-rating momentum (gradual one-way PE expansion likely backed by ETF passive buying) and perception of it being a premium sector with almost insurmountable moats ("once they're embedded in businesses' workflows & systems - almost impossible to walk away = sticky, annuity-type licensing revenue that deserves a much higher multiple"). Obviously both these factors have now been thrown into doubt or reverse - effectively placing a question mark over future revenue & terminal value assumptions - and hence the logical response to de-rate the PE multiple expansion. By how much is appropriate? ... now there's the million dollar question.

At least one of the most aggressive drivers of the rout has left the party (see below article from today's AFR), so your nibbling at Pro Medicus & other SaaS looks to have been fairly well timed....

Arnott Capital, Minotaur and Totus close their software shorts as rebound gathers steam

Yes, Mike, it's very difficult to make money out of this unless you are privy to flows and can see the extent of the traffic, and that's not me (and those that are can sound smart). it just a narrative that explains what's happened and maybe the force of the moves. For example, six months ago i wrote, i thought there were signs of a rotation coming after the August results. i thought it might be a high tide, not a tsunami, lol. otherwise i would have put everything into BHP and gone to the beach for 6 months!

i wonder if super intelligence AI (where AI does everything and most service companies are stuffed) can be added to the underlying equity risk that we continually live with, like thermo nuclear war or cannibalistic zombie pandemic, etc, a small % of happening but huge negative outcome, and we just have to know it exists, and get on with picking stocks.

@mikebrisy We cannot ruleout anything howver based on my recent visit to SF World Tour,I can see there is a narrtive change by Salesforce to project it self as system of record & trying to devlope its offering as a platfom where customer can pick & chooses services. At the enterprise level replacing salesforce seems to bea a far cry, the opportunties for vibe codig within the platfome to create salesforce platfome porducts in salesforce appexchange is enormous. the implementation timelines are getting shorter & less costlier as well- My version

@mikebrisy never say never, but the vibe at the SF World Tour was different this year. Salesforce is successfully rebranding as the System of Record, pivoting toward a "modular platform" model where customers curate their own suite of services.

Replacing Salesforce at scale? That's a mountain most enterprises won't climb. The real gold rush right now is in Vibe Coding within the ecosystem. The opportunity to build and launch AppExchange products is massive, especially as implementation cycles get faster and leaner.

The platform isn't just a database anymore; it’s a launchpad.

AI version

@Raseekingalpha "Replacing Salesforce at scale? That's a mountain most enterprises won't climb. "

The WSJ article (which I pasted below) makes the same point. Basically, it argues that the vulnerability is at the point in the SMB growth journey where they first decide to systematise their CRM processes, and want a way of doing things that incumbent software providers don't fully meet. So, once you have the data systems and workflows estbalished, and have started scaling, the switching cost is large.

I found the article interesting because Strawpeople(including me) have argued that an SMB owner would not build their own product, because they are too busy running their businesses - a view I am sympathetic too, and to do would likely be a poor allocation of resources. This underscores part of the SaaS thesis.

While the article provides some real examples that show that some SMB CEOs will develop and are developing their own solutions, these are of course isolated anecdotes and can be argued to be "edge" cases with perhaps unique CEOs, not representative of SMB CEOs more generally.

To the article below, I will add my own anecdote albeit not a CRM. For my international asset portfolio I retain an advisor who I meet twice a year and as needed. He is the principal in a small international wealth management firm. In our most recent call he told me that for many years, his firm has licenced a US will writing software program. They've paid about $US5,000/yr for the last 10 years. It is clunky, not user-friendly, but it is what they've had and used. Earlier this year, using Replet and Claude, in 4 days they spun up their own version of the software, validated it with the lawyers they partner with, and retired their licensed product. I ask him about risks, robustness etc. etc. and he said that ultimately, a human lawyer has to review the will, so they consider the risk is manageable. Its got rid of a horrible piece of software no-one liked, and has delivered a small, ongoing opex saving.

There are some attributes of this case example that show this is anything buy a typical SaaS example. The "Will Drafter" was not a system of record; there are limited security considerations; there are no integrations beyond (presumably) legal standards applicable to each jurisdiction which must be adhered to, or model form templates. But I found it interesting that my adviser had taken the effort to spin up a tool that was way outside their core competency, and that they had a working product in 4 days effort and have now terminated the legacy licence.

Was the business case there? Well, assume it was 4 days work at $US2,000/day so $US8,000 or a payback of < 2>

Again, it is just another (and admittedly weak) anecdote.

So, I remain open to the uncertainty and am alert to how this evolves. After all, my ASX portfolio is some 55% weighted to SaaS - indicating the buying I've been doing down into and through this dip!

Across all the more thoughtful commentary and analysis of this issue, several themes are emerging that identify why SaaS incumbents who effectively embrace AI might succeed: system of record; integrated standardised workflows; deep AI integration; security and reliability; embedded learning from customer feedback; ecosystem integration ("connections" and networks); customer core competency focus etc.

=========================

Meet the Companies Vibe Coding Their Own CRMs

Most large enterprises would be wary to walk away from an established CRM vendor like Salesforce. But for smaller companies with specific needs, it’s a different story.

By Isabelle Bousquette

March 4, 2026 11:00 am ET

Some small and midsize companies are vibe coding their own customer relationship management software to get more customized systems at a better price. KEVIN COOMBS/REUTERS

Quick Summary

• Small and midsize companies are using AI tools to build custom CRM software, displacing established vendors and causing investor concern.View more
• Predictions that AI tools could help displace established business software are quietly coming true in some segments of the market.
• A number of small and midsize companies say they are vibe coding their own customer relationship management software in an effort to get more customized systems at a better price. So-called CRMs are a critical business system for tracking, analyzing and taking action on sales, marketing and customer data, and it’s an area Salesforce has dominated for more than a decade.
• “We tried Salesforce and it was OK,” said Bill Schonbrun, chief operating officer of water treatment company CarboNet. “We did all kinds of customizations, but never got to the place where it did exactly what we wanted it to do.” Schonbrun ultimately opted to build his own custom CRM for the 65-person company with the help of AI. 
• For weeks, software stocks have borne the impact of investor concern that increasingly capable AI coding tools could help some companies displace behemoths with nimble, vibe-coded alternatives. Some of the largest software vendors, including Salesforce, tried to reassure panicked investors last week.

The market panic is overblown, said Stefan Slowinski, global head of software research at investment bank BNP Paribas. It is unlikely any complex multinational would suddenly walk away from a system as entrenched as Salesforce, he said. 

A Salesforce spokesperson said “vibe coding CRM without enterprise-grade protections for business data is a high-stakes gamble that [small and medium businesses] can’t risk,” citing a 2025 MIT research report that said internal builds were failing twice as often as strategic vendor partnerships.

Still, some small and middle market companies are forging ahead, finding with AI an opportunity to build systems that more closely fit their unique needs.

In January, Dave Clark, the former CEO of Amazon’s consumer operations and current CEO of supply chain startup Auger, went viral with a post about how he vibe coded a new CRM over the course of a weekend.

“We tried configuring an off-the-shelf tool for our cycle. Too many fields we don’t need, missing the ones we do…So I just built what we needed. Took a night and a morning,” he wrote. 

A Salesforce spokesperson said “vibe coding CRM without enterprise-grade protections for business data is a high-stakes gamble.” John G Mabanglo/EPA/Shutterstock

Critics were quick to point out these apps can often be easier to build than they are to scale, secure and maintain.

That’s why many companies who want to build their own enterprise-grade CRMs end up working with a vibe coding vendor, some of whom offer customizable CRM templates companies can build on, said Adnan Zijadic, a senior principal analyst at Gartner.

That was the case for Vancouver, British Columbia-based CarboNet, which worked with OverAI.

OverAI, a startup with five full time employees and $6.35 million in venture funding, offers a conversational “AI Architect,” that helps companies build their CRM, or any other given system in natural language, while also pulling in prebuilt components, like permissioning systems and security alerts, said co-founder and Chief Executive Mollie Breen.

Before working with OverAI, CarboNet’s Schonbrun said he tested different CRM providers, including Salesforce, but often found issues with their overhead and a user experience that was sometimes frustrating for salespeople.

He added he isn’t sure if the cost of building and maintaining a custom CRM—about $15,000 to $20,000 to build and $5,000 annually to maintain—was less than the cost of an enterprise SaaS license.

But the custom build offers a better user interface that the sales team actually uses, he said. 

“You can be incredibly specific with exactly what you want it to do. And that makes the user’s lives way easier,” he said. 

Whetstone Distribution, a meat processor and distributor in the Midwest, also worked with OverAI after determining it didn’t want software that would take a huge part of their budget, lock the firm in for a yearslong contract and require a lengthy implementation process.

So the firm custom built a tool that married sales and operations data in a visual map format for about $10,000, said CEO Beth LaBossiere. Building it out took a few weeks, with two to three people internally at Whetstone working with the OverAI team.

Crux Capital, an early-stage venture capital fund, also decided to build a custom CRM after struggling to find SaaS offerings that fit its needs. 

Venture-capital firms are a “unique animal,” said Lev Mass, a general partner at Crux. Their “customers,” aren’t customers in the traditional sense, but rather the startups they talk to in any given year as they consider potential investments. They are looking to identify trends over a long time horizon and stay in touch with companies that might be a good investment, even if it is sometimes years down the road.

Essentially, it is an extremely custom-use case, Mass said.

For the last few months, the firm has worked with startup AnySoft, which provides vibe-coding capabilities and some existing templates for the process. Mass said some portions of the CRM are now in production. 

The question of whether larger enterprises will seek to replicate some of these efforts remains to be seen, but it is top of mind for anxious investors.

In its earnings last week, Salesforce sought to reassure investors of its relevance in the AI age, reporting fourth quarter revenue rose 12% to $11.20 billion. Salesforce’s own AI product, Agentforce, reached $800 million in the quarter, up from $540 million the quarter before.

BNP Paribas’s Slowinski said some of the investor concern is less about customers walking away from established vendors in favor of vibe-coded alternatives and more that an explosion of new, AI-powered vendors could give customers negotiating leverage and put pricing pressure on the established players. 

Still, those who have built their own CRMs say large enterprises shouldn’t necessarily write off the possibility. 

“I think the Fortune 100 will begin to evolve over time,” said CarboNet’s ​​Schonbrun. “I think it’s a myth that big companies can’t or shouldn’t adopt this.”

@mikebrisy I’m finding this whole discussion fascinating, because the examples in the WSJ piece feel like early signals rather than the main event.

Right now, AI‑built internal tools still need human checking, and that naturally limits how far this can go. Most of the SMB cases look like simple, low‑risk systems where the downside is small and the oversight burden is manageable.

But the trajectory matters more than the anecdotes.

These tools are only at the very beginning of their curve. The direction of travel is toward systems that are easier to build, more stable, more secure, and eventually self‑validating. As the oversight burden drops, the economics change again — not just for SMBs, but potentially for larger organisations too.

I’m not suggesting enterprises will rebuild Salesforce tomorrow. But I do think we’re underestimating how quickly AI‑native systems will mature, and how that might shift the balance between “buy” and “build” over the next few years.

For now, the moats around system‑of‑record platforms still look strong. I just wouldn’t assume the current limitations of AI‑built software are permanent.

@Foxlowe I agree. With perhaps one adjustment.

I think we might not be underestimating the pace of change, but we might be under-estimating the EXTENT of the change.

However, I agree with you, in that how we see the drivers of "buy" and "build" over the next few years will evolve significantly.

That is a key lesson identified in "The Innovator's Dilemma" (Clayton Christiansen; 1997).

I was having a conversation earlier this week with some friends in my age group (55-60). ... ( like the joke goes,... a Kenyan, an Icelander and a Kiwi walk into a bar...)

We were reflecting on some of our practices in the 1980s and 1990s ... and what it was like to live then ... tech and more. I don't think we could have written the script for 2026.

I like the way you’ve framed that, @mikebrisy - not just the pace of change, but the extent of it. That’s the part that’s hardest to get our heads around, because we’re all anchored to the world we grew up in.

Your reference to The Innovator’s Dilemma is spot on. Disruption rarely looks dangerous at the start. It begins at the fringes with tools that seem too small, too simple, or too “toy‑like” to matter. Then the capability curve steepens, and suddenly the boundary between “buy” and “build” doesn’t look the same anymore.

That’s the piece I keep coming back to with AI‑built internal systems. Today they still need human checking, and that naturally limits their scope. But that oversight burden won’t be permanent. As these tools become more stable, more secure, and more self‑validating, the economics shift again — and probably further than we expect.

It also reminds me of the early days of computing. IBM famously said the world would only ever need one computer. Then they revised it to three — one for America, one for Europe, one for Asia. And look where we are now. Most households have half a dozen devices, and computers are embedded in everything from fridges to cars to doorbells.

The pattern is the same: once a technology becomes easier to build, cheaper to run, and more reliable, it doesn’t just scale — it saturates. The early predictions always look conservative in hindsight.

Which is why I’m cautious about assuming today’s limitations on AI‑native systems will hold for long. We’re still at the “one computer for the world” stage of this curve.

Just a quickie:

That's a chart from Marcus Padley's MarcusToday Saturday weekly wrap email today, and it shows the movements in the major global exchanges and indices over this past week; isn't it interesting that when you take the "Magnificent 7" out of the S&P500, the remaining S&P493 was actually NOT in the red, but was up 0.40%?! Yet the S&P500 was down -2.02%, so (1) that really highlights how much of an influence the Mag 7 has on the S&P500, and (2) Software and IT in general is still on the nose.

Those moves across European exchanges (the bottom 6 red lines) make our sell-off look quite tame by comparison.

We are certainly living in interesting times.

For any newbies who are unaware, the Magnificent 7 stocks - Apple (AAPL), Microsoft (MSFT), Alphabet (GOOGL), Amazon (AMZN), Meta Platforms (META), Nvidia (NVDA), and Tesla (TSLA) - are in both the S&P 500 and the NASDAQ Composite (and specifically the NASDAQ-100). As some of the largest companies in the world, they are foundational members of the S&P 500 and simultaneously represent a massive portion of the tech-heavy NASDAQ. According to Google just now, the Information Technology (IT) sector constitutes approximately 32% of the total value of the S&P 500 and about 40% to 45% of the NASDAQ-100. By contrast, here in Australia our IT sector is only around 2% of the ASX200 Index.