Bit of a mammoth issue for you today, given I was out for a few weeks. In my defense, those few weeks involved: wrapping up a very demanding job, driving 18 hours for the family to be on vacation, starting and onboarding to a new job, attempting to drive back, an incident with a deer, a busted RV that had to be towed in another state, a much smaller rental van, immediately going back to work after another 18 hour-drive home, and now a sick dog.

So like. I feel that I earned those clip episodes.

We revisited my warning about irresponsibly deploying customer-facing AI, and looked back at some of the most infamous AI jobs to have gone through Bad Job Bingo.

I relaunched my personal website with an elegy for the old internet, and also reminded myself not to be a little shit.

Okay, maybe not everyone. But they’ve certainly been busy: since June 1, Salesforce has signed definitive agreements to acquire Contentful, m3ter,¹ and the big one, Fin (formerly Intercom).

I call Fin the big one because Salesforce will pay an eye-watering $3.6 billion for the company, which rebranded in May as it pivoted hard from being mainly a ticketing platform to being all about its full-service AI customer agent. For comparison, Zendesk acquired a similar company, Forethought, in March in a reportedly nine-figure deal.

The difference in valuation here is interesting to me because, while Forethought has been a customer AI and automation company since its founding in 2017, Fin went all in on the AI game relatively recently. Which is not to say it’s inferior (honestly, I wouldn’t know, I haven’t used either product), but I will say that I also find the timing of these deals interesting.

I’ve been pretty open about my skepticism regarding the future of AI, and I’m not the only one. Ed Zitron, whom I’ve just linked, has written repeatedly and extensively on the economics of AI or, as he has aptly termed it, the brokenomics:

Basically, either AI companies’ customers will go broke trying to pay for their AI tokens when the AI companies eventually start charging the real rate for usage, or — more likely — customers will give up on AI in enough numbers to bankrupt the entire AI ecosystem. And it won’t matter whether CX companies rely on external AI technology or just on their customers’ ability to invest in their AI products, because it’s bad news either way.

Although Fin’s Operator product still runs on Anthropic’s Claude, its agentic capabilities are built on its own proprietary AI model (Apex 1.0). Salesforce has made some controversial choices recently in an effort to cut costs and retool its operations around AI. Its share price has fallen 41.3% in the past year, and its returns are down 36.2% over the past 5 years, so only time will tell whether it can sustain the kind of investment a proprietary AI model requires (given that the major players are struggling despite having many, many more billions of dollars than Salesforce, I’m doubtful).

Zendesk has had a similar trajectory, and although it claims its recurring AI revenue could hit $500 million this year, Forethought’s SupportGPT is built on OpenAI’s models,² making it entirely dependent on external technology.

But let’s say Salesforce can commit that kind of cash and Zendesk’s revenue is indeed on the rise. Even then, if the AI bubble bursts? Well, Zitron said it: everything will begin to fall apart.

I’m not sure what resources Salesforce or Zendesk will be able to muster if its customers, its investors, and the companies behind the technology — basically the entire AI infrastructure and a frighteningly large chunk of the Tech industry — are all broke.

I’ve seen some in our industry point to Fin’s acquisition as an example that AI in customer service is a sure bet. For one example, here’s Kustomer CEO Brad Birnbaum's take on it:

I’m going to be frank here: given the actual state of the major players’ finances and the fact that the AI industry is basically just passing “earnings” and investment funds among itself, that’s either an extraordinarily naive understanding of reality or an extraordinarily dishonest one.

This purchase doesn't prove anything positive about the viability of AI in customer service, or the future of CX platforms. If anything, it demonstrates that the Tech industry in general, and the CX sector specifically, can read the writing on the wall.

The whales are desperate for their investments in AI to pay out somehow before the bubble bursts, and the little fish — like CX platforms who made AI their entire identity — are desperate to cash out before the same.

I mean, look. I hope I’m wrong. I hope we’re all wrong. I don’t think we are, but it would be nice if I didn’t live through another financial collapse in my lifetime. But I don’t think ignoring the conditions that could make it happen will somehow prevent it from happening.

We should use the time we have to prepare for that increasingly likely eventuality. As CX leaders, that means discouraging our companies from replacing vital human resources and operations with AI tools they won’t be able to afford (or even expect to exist) in a year or two. Because once that expertise and institutional knowledge are gone, there’s no telling if they’ll have the means to get it back.

But knowing our industry as I do…that’s perhaps an unrealistic expectation. And I know it is just as unrealistic to expect executive teams to be honest about how shaky their strategies and investments are when they can make a lot of money by fooling people instead, then use their money to make a literal cash parachute and bail out before the AI rocket falls back to earth.

I just want it on the record that that’s what they’re doing.

I think this is the most people we’ve ever celebrated in a Roundup, which makes me so happy because this section is like 85% of the reason I do this newsletter. Y’all are amazing, and thank you for trusting me to boost you!

Paresh Dave wrote a piece titled ‘Tell Him He’s a Piece of Shit’ on the poor morale at Meta (the 'him' being an exec at Meta). The opening anecdote to this story was wild and then it just gets wilder. I don’t think I’ve ever heard an executive acknowledge shitty working conditions so plainly. I’m not sure it's possible to right the ship as long as Zuckerberg is still in charge, but I’ll be watching with popcorn either way.

Afshan Kinder wrote about integrating the right technology at the right time.

Claudia R. shared a story about a very (and probably unintentionally) funny interaction they once had with an exec.

Lara Ballard wrote about why privacy professionals should care about post-quantum cryptography.

Ines van Dijk wrote about what happens when we take the decomposition vs deflection problem seriously. This is so good. SO GOOD. I had a really hard time picking just one quote, and the final line is a mic drop. I can't wait to read Ines' paper when it comes out.

Sarah Betts shared how she built a powerful network. Really fantastic advice here, and I echo it. I always say the fastest way to network (which is really just getting to know people) is to just...be a person. Shoot the breeze. Talk about your favorite book/tv show/video game. Don't try to anticipate what fleeting professional thing people will be interested in at any given moment; if there's a natural opening to show your chops, take it. But otherwise just let moments breathe.

Sarah also wrote a fantastic piece on why the CSAT is not the customer, with an opening graphic that made me actually cackle. (Also another one where I had a tough time picking just one quote. Always a great measure of quality for me, lol.)

Jillian Bejtlich wrote about what you promised when you first built your community (even if you didn't know it).

Samantha O'Neil wrote about how she went from a chaos coordinator to thriving in a healthy job.

Chrissy Sebald asked whether we're making things better or worse by introducing AI. Woof, I was feeling my blood pressure rise with every sentence reading this. As Brian Levine and Miles Goldstein noted in the comments on the post, much of what Chrissy described isn't new tech; there are just companies out there that don't care enough to bring old and new tech together to do the best for their customers.

Gillian Wu gave us the 10 Commandments of Community Management.

Rob Dwyer wrote about how events like CCW provide an opportunity for anyone involved in CX to be present with members of their community.

James Baldwin wrote about the difference between the AI you play with and the AI you bet on. James provides incredibly useful advice for how to get your AI tool to do what you need it to do reliably. And when I say useful, I mean real, nitty-gritty, in-depth steps for conducting effective testing. Great stuff. (Speaking of reliability, James is consistently a great resource for this kind of thing, so he's worth a follow if you're looking for real talk on AI.)

Kate Nelson wrote about how, often, the most memorable customer experience is simply someone who makes you smile when you need it most.

Dylan Bates wrote about how, when you're new to Success, the fastest progress comes from learning the customer’s path through the product.

The Table Service Podcast interviewed Justin Robbins about how the real work comes after the motto.

Support Stack talked to Dawn Perrot about the feature that changed her role: Fin's Operator memory.

Alli K. turned our corporate experiences into the Oregon Trail parody we didn't know we needed.

Rexxfield Investigates spoke with Dave Pilot about how banks, law enforcement, and threat intelligence teams fight modern scams. Ronnie is a friend who works in the fraud/scam prevention and response space. I especially liked this episode of his podcast given the focus on what communities can do to protect elders, families, and kids online.

Bit hot ain't it.

The Supportive spoke with Pavel Malyshev about his honest, insider’s view of AI in customer support jobs.

Unresolved.cx talked to Yvette Johns about the development conversation that managers keep skipping for themselves. This is such a great episode! Super entertaining, and Yvette kills it with her wisdom on developing oneself and others (she is both a clear expert and super relatable). This is unrelated, but I also really love how Brett presents each episode of the podcast—showcasing his guests and breaking down their key takeaways. It shows the care he has for his guests and his listeners.

CX Now talked to Neil Smith about how elite tech support teams protect customer trust.

Crash Out With Me spoke with Taryn Talley about how hiring feels like a humiliating ritual. A timely discussion given the job market CX has been experiencing. It's hard to reconcile the super shitty job search and interview experience with the fact that, anecdotally at least, I've seen hiring pick up for CX roles in the last few months. I think the driving factors are actually pretty complicated, but in an over-simplified nutshell: IMO, it's a strange combination of the terrible economy (which shifts power from workers to companies) and reality setting in about AI (it's not actually gonna replace everyone).

The Not-Boring Tech Writer talked to Vladimir Izmalkov about mapping your technical writing skill tree. They had me at "RPG-inspired skill tree."

The Agentic Shift: Navigating the New Human-AI Workflow
July 13-14 in NYC, NY. Hosted by All Tech is Human and Tech Ethics Lab. Register here.

Agentic Customer Support Summit
July 15 at 9am ET. Hosted by Customer Success Collective. Register here.

Community as Infrastructure & Office Hours That Drive Adoption
July 15 at 10am PT. Webinar hosted by CMX by Bevy. Register here.

Quantum Frontiers & Responsible AI: A Global Dialogue on Emerging Technologies
July 15 at 3pm ET, NYC, NY. Hosted by All Tech is Human. Register here.

How to Transition Into Responsible Tech: Career Pivot Focus
July 16 at 10am PT. Webinar hosted by All Tech is Human. Register here.

How customer success can define, achieve, and prove value
July 16 at 10am PT. Webinar hosted by Customer Success Collective. Register here.

TrustCon 2026
July 20-22 at Hyatt Regency, San Francisco, CA. Hosted by TSPA. Call for proposals open now.

Community and AI: A Fireside Chat
July 21 at 9am PT. Hosted by CMX by Bevy. Register here.

Scaling CS with AI and automation
July 21 at 12pm ET. Webinar hosted by Customer Success Collective. Get tickets here.

The Neurodiversity Gap in Experience Design
July 30 at 11am ET. Webinar hosted by CXPA. Get tickets here.