Can someone track down the gate agent who didn’t let @balogun mom on the flight back to England cause she looked too pregnant? After last nights performance they deserve front row seats to every @USMNT match this World Cup.

Best half of soccer I’ve ever seen from a @USMNT and I’ve watched a ton of us soccer

I just went 74-8. How is this squad not going 82-0? 82-0.com/share?id=LIy4U4MDY0… #NBA #82and0

Kustomer Architect Aims to Replace CX Deflection Metrics With Business Outcomes cmswire.com/customer-experie…

For all the AI will replace all the customer service jobs, riddle me this

Another example of why I agree with @DavidSacks on AI driven job creation, not @Jason’s prediction that it’ll be net negative.

I have been cold calling for nearly 20 years. That is a lot of accumulated knowledge on what works and what does not. What if you could clone my brain and produce my best cold outreach in seconds? That is basically what I have been able to do after Kustomer doubled down on our investment in Claude. I have been at Kustomer for nearly 7 years. In that time, a big part of my pipeline has come from cold outreach. The challenge is that the tactics that actually work are not scalable. Writing something thoughtful, relevant, and worth responding to takes time. And pipeline is the lifeblood of any sales org, especially right now. Earlier this year, I ran a training for my team and shared examples of messaging that has worked for me. It helped, but it was not durable. It still required a lot of effort to replicate. That changed this weekend. I went deep on Claude and built a custom skill in Claude Code. I fed it years of my highest performing emails and LinkedIn messages. I also fed it a number of my LinkedIn posts about Kustomer’s value so it could learn how I communicate our messaging. I layered in existing skills created for Kustomer positioning, case studies, and marketing content. I connected Salesforce data to understand current customers and history. I also pulled in public signals like press, podcasts, and posts. Now I can drop in a prospect’s LinkedIn profile and it identifies the same triggers I have historically used, synthesizes relevant context, and writes a hyper personalized message in my voice. What used to take 30 plus minutes now takes under a minute. Most of the time, the first draft is strong. This is AI done right. Not cold outreach slop (I get 50 of those a day), but something that captures what takes someone's superpower and scales it. Next step is layering in the Claude in Chrome extension which will be able to pull common connections and read through LinkedIn posts pushing personalization even further. This skill is now made available for my whole team. No more stale trainings but an actionable tool that can be used to drive immediate value. Exciting times.

If you follow AI conversations, you’ve probably heard “human in the loop.” In theory, it means a human can step in, review, approve, or take over when the AI falls short. It’s supposed to reduce hallucinations and improve outcomes. But in CX right now, I’m seeing the opposite: human out of the loop. Two big reasons: 1) Disconnected systems Most bolt-on AI agents aren’t deeply integrated with where humans actually work. The AI runs its course, then hands off a ticket. Now the human has to reverse-engineer what just happened. By the time they catch up, the customer’s already moved on. 2) Asymmetry in tools and knowledge Your AI agent might be hooked into APIs, internal systems, and rich knowledge sources. But when it fails? The human gets a stripped-down ticketing system with none of that context. So the handoff doesn’t just slow things down, it guarantees another failure. We’ve all felt this: You interact with an AI agent, don’t get the answer, get passed to a human… and have to repeat everything. That’s not “human in the loop.” That’s broken. Context is the unlock. Without a complete view of the customer journey, across both AI and human, you can’t deliver a seamless experience. Most tools today fall short here. At Kustomer, this is exactly what we’re built to solve.

For those trying to return their @Allbirds for AI compute, unfortunately they haven’t included this a refund option

Looks like @claudeai is down right now. Luckily, they can buy more compute from @Allbirds

If seeking pain is what makes a great startup employee @Alfred_Lin, consider me the @davidgoggins of software sales

Last week I had a great conversation with my friend who’s been in the BPO space for 15 years. A few trends from our discussion stood out. The AI headcount narrative doesn’t match what he’s seeing. Based on conversations across his BPO network, Taylor isn’t seeing the mass support layoffs the AI headlines suggest. AI ROI calculations often miss key variables. When you add AI to your site, you’re often introducing an entirely new channel. That frequently increases inbound volume by 30% , which must be factored into the ROI. Many CX leaders are being asked to deliver outcomes tied to AI hype cycles. That can put them in a difficult position when expectations don’t match operational reality. “Resolution rates” don’t tell the full story. Metrics often miss channel switching to human agents, drop-off, or silent churn. What’s the hidden cost of a frustrated customer who has a bad AI interaction and simply never comes back? Jevons Paradox will likely apply to CX. I’ve never spoken with a company that thinks they’re delivering too much great customer experience. Most want to do more, but cost is the constraint. If AI reduces Tier 1 issues, my bet is we’ll see more human support focused on higher-value interactions, not less.

Proud to BU

When hiring sales reps, I think about three buckets. Knowledge, Skill and Will. Knowledge, your understanding of our space, buyer, product, competitors, etc. Skill, your ability to prospect, multi-thread, forecast, negotiate, etc. Will, your ability to handle change, deal with conflict, be resourceful, have a growth mindset. Over index on the Will bucket. Skill and Knowledge can be taught. Will can’t.

Personal update: I’m launching my own podcast. It’s called Founding Sellers, and the idea is simple. I’ll be interviewing the best early stage GTM operators and IC's who help scale a product from zero to sixty. For those who know me well, you know I’m obsessed with podcasts. Top 1% of listeners on Spotify. Any moment of free time, at the gym, walking the dog, doing chores around the house, driving, I’m usually listening at 1.25x. Sometimes 1.5x if I’m feeling dangerous. It’s how I feed my obsession with learning, and it’s often the spark for my best ideas. A couple months ago at a team dinner, we asked each other what our ideal career would be if money didn’t matter. I said podcasting. My team looked at me like I was crazy. Eventually one of them said, “You know you can do that right now.” They were right. The only things getting in my way were fear and excuses. One of my themes for the year is getting comfortable being uncomfortable. So here we are. The goal of the podcast is to share lessons from people who have actually been in the arena. I’ve been lucky to be an early stage seller at three startups that eventually successfully exited to Zoom, Medallia, and Meta. Most podcasts focus on founders or VCs. I love that content, but it often misses what’s happening on the front lines. What actually works in the early days. What doesn’t. And what it takes to survive and succeed as a founding seller. The first episode with my good friend @TheRealMcHoy , early stage seller at Pendo.io and founder of Atlas, drops next week. In the meantime, enjoy this clip. And if you know someone who would be a great guest, submit them on the site or send me a DM.

At our company kickoff last week, we talked about how speed is the only moat left in technology. As the cost of building software trends toward zero, the barrier to entry disappears. You don’t need to be a trained engineer anymore. Even a guy with 15 years in sales and zero coding education. (Aka me.) This weekend I saw it firsthand. It’s territory season in sales. Historically, ops teams rely on enrichment tools like ZoomInfo, Clearbit, BuiltWith, etc. The problem? The data is often outdated. We’re targeting a specific competitor, so ops pulled a list of accounts tagged with that technology. When I started reviewing the accounts, a huge percentage were actually using competitive products. In the past, I’d either: 1.Spend hours manually validating 2.Or accept it as a limitation of scaling a sales org Not anymore. I opened up Replit. I gave it a prompt explaining the situation. Fed it 10 examples of sites I knew were using the tech. Had it identify variations of the relevant code snippets and confirming signals. Less than an hour later, I had a working prototype where I could: •Paste a URL to validate •Bulk upload URLs •Confirm whether they’re actually using the provider Over the last two days, my ops leader and I have been refining bulk uploads and export functionality. Total cost: $50 on Replit. Total time: a few hours. Outcome: A tool that saves AEs hours of manual work and allows marketing to more precisely target a competitor. If I can build this, anyone can. The only difference now is who moves. Start building. Control your own leverage.

What a succesful AI deployment in CX looks like. One that balances automation, efficiency and customer delight. CSAT up 40% since implementing. markets.businessinsider.com/…

Green “Biscayne” Bay